From fd77d007ea4d8a715db77cb1e8f78ca11cf9b548 Mon Sep 17 00:00:00 2001
From: Todd Morrill <tmorrill12@gmail.com>
Date: Fri, 22 May 2020 19:26:38 -0400
Subject: [PATCH 1/6] initial skorch hyperparam opt implementation

---
 .gitignore                                   |    4 +
 machine-learning/model.py                    |   44 +
 machine-learning/skorch-hyperparam-opt.ipynb | 1117 ++++++++++++++++++
 3 files changed, 1165 insertions(+)
 create mode 100644 machine-learning/model.py
 create mode 100644 machine-learning/skorch-hyperparam-opt.ipynb

diff --git a/.gitignore b/.gitignore
index 70702445..70255b20 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,3 +4,7 @@ _build/
 mydask.png
 dataframes/data
 .idea/
+.devcontainer/
+.data/
+.vector_cache/
+__pycache__
\ No newline at end of file
diff --git a/machine-learning/model.py b/machine-learning/model.py
new file mode 100644
index 00000000..54cc9192
--- /dev/null
+++ b/machine-learning/model.py
@@ -0,0 +1,44 @@
+# more details can be found here: https://github.com/bentrevett/pytorch-sentiment-analysis/blob/master/4%20-%20Convolutional%20Sentiment%20Analysis.ipynb
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class CNN(nn.Module):
+    def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.2, pretrained_embeddings=None):
+        
+        super().__init__()
+        self.embedding = nn.Embedding.from_pretrained(torch.FloatTensor(pretrained_embeddings))
+        self.embedding.weight.requires_grad = False # save some computation
+        embedding_dim = self.embedding.embedding_dim
+        self.conv_0 = nn.Conv1d(in_channels = 1, 
+                                out_channels = n_filters, 
+                                kernel_size = (filter_sizes[0], embedding_dim))
+        self.conv_1 = nn.Conv1d(in_channels = 1, 
+                                out_channels = n_filters, 
+                                kernel_size = (filter_sizes[1], embedding_dim))
+        self.conv_2 = nn.Conv1d(in_channels = 1, 
+                                out_channels = n_filters, 
+                                kernel_size = (filter_sizes[2], embedding_dim))
+        self.fc = nn.Linear(len(filter_sizes) * n_filters, 2)
+        self.dropout = nn.Dropout(dropout)
+        
+    def forward(self, text):
+        #text = [batch size, sent len]
+        embedded = self.embedding(text)
+        #embedded = [batch size, sent len, emb dim]
+        embedded = embedded.unsqueeze(1)
+        #embedded = [batch size, 1, sent len, emb dim]
+        conved_0 = F.relu(self.conv_0(embedded).squeeze(3))
+        conved_1 = F.relu(self.conv_1(embedded).squeeze(3))
+        conved_2 = F.relu(self.conv_2(embedded).squeeze(3))
+        #conved_n = [batch size, n_filters, sent len - filter_sizes[n] + 1]
+        pooled_0 = F.max_pool1d(conved_0, conved_0.shape[2]).squeeze(2)
+        pooled_1 = F.max_pool1d(conved_1, conved_1.shape[2]).squeeze(2)
+        pooled_2 = F.max_pool1d(conved_2, conved_2.shape[2]).squeeze(2)
+        #pooled_n = [batch size, n_filters]
+        cat = self.dropout(torch.cat((pooled_0, pooled_1, pooled_2), dim = 1))
+        #cat = [batch size, n_filters * len(filter_sizes)]
+        logits = self.fc(cat)
+        #logits = [batch_size, output_dim]
+        return F.softmax(logits, dim=-1)
\ No newline at end of file
diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
new file mode 100644
index 00000000..54344473
--- /dev/null
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -0,0 +1,1117 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install -q dask_cuda torch torchtext skorch\n",
+    "!pip -q install dask[dataframe] --upgrade"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Hyperparameter optimization with Skorch\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Setup Dask"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<table style=\"border: 2px solid white;\">\n",
+       "<tr>\n",
+       "<td style=\"vertical-align: top; border: 0px solid white\">\n",
+       "<h3 style=\"text-align: left;\">Client</h3>\n",
+       "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42667</li>\n",
+       "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
+       "</ul>\n",
+       "</td>\n",
+       "<td style=\"vertical-align: top; border: 0px solid white\">\n",
+       "<h3 style=\"text-align: left;\">Cluster</h3>\n",
+       "<ul style=\"text-align: left; list-style:none; margin: 0; padding: 0;\">\n",
+       "  <li><b>Workers: </b>1</li>\n",
+       "  <li><b>Cores: </b>1</li>\n",
+       "  <li><b>Memory: </b>33.68 GB</li>\n",
+       "</ul>\n",
+       "</td>\n",
+       "</tr>\n",
+       "</table>"
+      ],
+      "text/plain": [
+       "<Client: 'tcp://127.0.0.1:42667' processes=1 threads=1, memory=33.68 GB>"
+      ]
+     },
+     "execution_count": 2,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "import torch\n",
+    "from dask_cuda import LocalCUDACluster\n",
+    "from distributed import Client\n",
+    "\n",
+    "# if you have GPU(s), use dask_cuda to automatically make use of them in your dask cluster\n",
+    "if torch.cuda.is_available():\n",
+    "    cluster = LocalCUDACluster()\n",
+    "    client = Client(cluster)\n",
+    "else:\n",
+    "    client = Client(processes=False, threads_per_worker=4,\n",
+    "                    n_workers=1, memory_limit='2GB')\n",
+    "client"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# for reproducibility\n",
+    "import random\n",
+    "import numpy as np\n",
+    "\n",
+    "SEED = 42\n",
+    "\n",
+    "random.seed(SEED)\n",
+    "np.random.seed(SEED)\n",
+    "torch.manual_seed(SEED)\n",
+    "torch.backends.cudnn.deterministic = True"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Create Data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torchtext\n",
+    "from torchtext import data\n",
+    "from torchtext import datasets"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# takes approx. 10 minutes to download data and embeddings (will be cached for re-use)\n",
+    "\n",
+    "# set up fields\n",
+    "TEXT = data.Field(lower=True, include_lengths=True, batch_first=True)\n",
+    "LABEL = data.Field(sequential=False, unk_token=None)\n",
+    "\n",
+    "# make splits for data\n",
+    "train, test = datasets.IMDB.splits(TEXT, LABEL)\n",
+    "\n",
+    "# will be used to initialize model embeddings layer\n",
+    "vocab = torchtext.vocab.GloVe(name='6B', dim=100)\n",
+    "\n",
+    "# build the vocabulary\n",
+    "max_size = 25_000 # shorten for demonstrative purposes\n",
+    "TEXT.build_vocab(train, vectors=vocab, max_size=max_size)\n",
+    "LABEL.build_vocab(train)\n",
+    "\n",
+    "# make iterator for splits\n",
+    "train_iter, test_iter = data.BucketIterator.splits(\n",
+    "    (train, test), batch_sizes=(32, 64), device=device)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['<unk>', '<pad>', 'the', 'a', 'and']"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# itos := index-to-string\n",
+    "# note the 2 extra tokens added for us: '<unk>', '<pad>'\n",
+    "TEXT.vocab.itos[:5]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "assert (len(TEXT.vocab.itos) == max_size + 2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.']\n",
+      "\n",
+      "pos\n"
+     ]
+    }
+   ],
+   "source": [
+    "# peek at the data\n",
+    "print(train.examples[0].text)\n",
+    "print()\n",
+    "print(train.examples[0].label)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# sadly train_iter isn't actually an iter..\n",
+    "# peek at a batch of data\n",
+    "batch = next(iter(train_iter))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor([[  45,  123,   25,  ...,    1,    1,    1],\n",
+      "        [5311, 3868,    2,  ...,    1,    1,    1],\n",
+      "        [ 133,    8,   42,  ...,    1,    1,    1],\n",
+      "        ...,\n",
+      "        [   9,  176,   10,  ...,    1,    1,    1],\n",
+      "        [1467,   10,   30,  ...,    1,    1,    1],\n",
+      "        [  82, 1297,    9,  ...,    1,    1,    1]], device='cuda:0')\n",
+      "tensor([147, 347, 361,  66, 147, 262, 336, 105, 255, 132, 128, 162, 122, 184,\n",
+      "        195, 100, 226, 127, 152, 159, 120, 204, 267, 209, 129, 486,  95, 151,\n",
+      "        140, 247, 127, 167], device='cuda:0')\n"
+     ]
+    }
+   ],
+   "source": [
+    "# numericalized tokens\n",
+    "print(batch.text[0])\n",
+    "# sequence lengths\n",
+    "print(batch.text[1])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "tensor([1, 0, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0,\n",
+       "        1, 1, 1, 1, 0, 1, 0, 0], device='cuda:0')"
+      ]
+     },
+     "execution_count": 12,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "batch.label"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "defaultdict(None, {'neg': 0, 'pos': 1})"
+      ]
+     },
+     "execution_count": 13,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# stoi := string-to-index\n",
+    "# check on the meaning of these zeroes and ones\n",
+    "LABEL.vocab.stoi"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Define your network"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# was having trouble with when model was defined in the notebook\n",
+    "# Can't get attribute ‘CNN' on <module ‘__main__'\n",
+    "from model import CNN"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "tensor([[0.4709, 0.5291],\n",
+       "        [0.2993, 0.7007],\n",
+       "        [0.2609, 0.7391],\n",
+       "        [0.3476, 0.6524],\n",
+       "        [0.4323, 0.5677],\n",
+       "        [0.3394, 0.6606],\n",
+       "        [0.2745, 0.7255],\n",
+       "        [0.3152, 0.6848],\n",
+       "        [0.4568, 0.5432],\n",
+       "        [0.4189, 0.5811],\n",
+       "        [0.4351, 0.5649],\n",
+       "        [0.4204, 0.5796],\n",
+       "        [0.3963, 0.6037],\n",
+       "        [0.3482, 0.6518],\n",
+       "        [0.3528, 0.6472],\n",
+       "        [0.4137, 0.5863],\n",
+       "        [0.3327, 0.6673],\n",
+       "        [0.3380, 0.6620],\n",
+       "        [0.4874, 0.5126],\n",
+       "        [0.2904, 0.7096],\n",
+       "        [0.2868, 0.7132],\n",
+       "        [0.3196, 0.6804],\n",
+       "        [0.3689, 0.6311],\n",
+       "        [0.2525, 0.7475],\n",
+       "        [0.2493, 0.7507],\n",
+       "        [0.3308, 0.6692],\n",
+       "        [0.2088, 0.7912],\n",
+       "        [0.4834, 0.5166],\n",
+       "        [0.2363, 0.7637],\n",
+       "        [0.4511, 0.5489],\n",
+       "        [0.2282, 0.7718],\n",
+       "        [0.3502, 0.6498]], device='cuda:0', grad_fn=<SoftmaxBackward>)"
+      ]
+     },
+     "execution_count": 15,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# smoketest\n",
+    "model = CNN(pretrained_embeddings=vocab.vectors).to(device)\n",
+    "model(batch.text[0].to(device))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "del model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Quick attempt at model training"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import skorch\n",
+    "from skorch import NeuralNetClassifier\n",
+    "import torch.optim as optim"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# this is a really unfortunate hack to make torchtext batching semantics work with skorch and dask\n",
+    "# the downside here is that we're no longer padding to the longest sequence in the batch, rather\n",
+    "# we're padding to the longest sequence in the *dataset*, which results in signifcantly more\n",
+    "# computation and thus significantly more time to train a model\n",
+    "# of course, you could set a max sequence length but that's not an ideal solution\n",
+    "# another solution would be to create a different dataset object, but then you can't use torchtext,\n",
+    "# which really is quite handy\n",
+    "\n",
+    "# train=True shuffles the data\n",
+    "train_iter_skorch = torchtext.data.Iterator(train, batch_size=len(train), train=True, sort=False, device='cpu')\n",
+    "test_iter_skorch = torchtext.data.Iterator(test, batch_size=len(test), train=False, sort=False, device='cpu')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# takes some time to numericalize the whole dataset\n",
+    "\n",
+    "# also notice that skorch and dask expect numpy arrays, which isn't ideal since it ties you to the cpu.\n",
+    "# meanwhile, projects like https://rapids.ai/ are moving toward all GPU computation, avoiding the cpu altogether.\n",
+    "for batch in train_iter_skorch:\n",
+    "    X_train = batch.text[0].numpy()\n",
+    "    y_train = batch.label.numpy()\n",
+    "\n",
+    "for batch in test_iter_skorch:\n",
+    "    X_test = batch.text[0].numpy()\n",
+    "    y_test = batch.label.numpy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(25000, 2470)"
+      ]
+     },
+     "execution_count": 20,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# notice how awfully large the second dimension is\n",
+    "X_train.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch.nn as nn"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# NB: not ideal to be using softmax + log + NLLLoss\n",
+    "# see discussion: https://github.com/skorch-dev/skorch/issues/637\n",
+    "skorch_model = NeuralNetClassifier(\n",
+    "                CNN,\n",
+    "                device=device,\n",
+    "                max_epochs=2,\n",
+    "                lr=0.001,\n",
+    "                optimizer=optim.Adam,\n",
+    "                criterion=nn.NLLLoss,\n",
+    "                train_split=skorch.dataset.CVSplit(.2), # NB: this witholds 20% of the training data for validation\n",
+    "                module__n_filters=100,\n",
+    "                module__filter_sizes=(2,3,4),\n",
+    "                module__dropout=0.2,\n",
+    "                module__pretrained_embeddings=vocab.vectors,\n",
+    "                batch_size=32,\n",
+    "                verbose=2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "  epoch    train_loss    valid_acc    valid_loss      dur\n",
+      "-------  ------------  -----------  ------------  -------\n",
+      "      1        \u001b[36m0.6302\u001b[0m       \u001b[32m0.7516\u001b[0m        \u001b[35m0.5243\u001b[0m  66.5468\n",
+      "      2        \u001b[36m0.4797\u001b[0m       \u001b[32m0.7994\u001b[0m        \u001b[35m0.4413\u001b[0m  59.9177\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
+       "  module_=CNN(\n",
+       "    (embedding): Embedding(400000, 100)\n",
+       "    (conv_0): Conv1d(1, 100, kernel_size=(2, 100), stride=(1,))\n",
+       "    (conv_1): Conv1d(1, 100, kernel_size=(3, 100), stride=(1,))\n",
+       "    (conv_2): Conv1d(1, 100, kernel_size=(4, 100), stride=(1,))\n",
+       "    (fc): Linear(in_features=300, out_features=2, bias=True)\n",
+       "    (dropout): Dropout(p=0.2, inplace=False)\n",
+       "  ),\n",
+       ")"
+      ]
+     },
+     "execution_count": 23,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "skorch_model.fit(X_train, y_train)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "0.79112"
+      ]
+     },
+     "execution_count": 24,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# quick check on the test set accuracy\n",
+    "skorch_model.score(X_test, y_test)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(array([0, 1]), array([12500, 12500]))"
+      ]
+     },
+     "execution_count": 25,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# random guessing would 50% accuracy so the model is indeed training\n",
+    "np.unique(y_test, return_counts=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "del skorch_model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Grid search with Hyperband"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# https://ml.dask.org/hyper-parameter-search.html#hyperband-parameters-rule-of-thumb\n",
+    "EPOCHS = 5\n",
+    "NUM_TRAINING_EXAMPLES = len(train)*.8\n",
+    "n_examples = EPOCHS * NUM_TRAINING_EXAMPLES\n",
+    "n_params = 8\n",
+    "\n",
+    "# it's not immediately obvious to beginners how all these parameters interact with each other\n",
+    "max_iter = n_params\n",
+    "chunk_size = n_examples // n_params"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# suppose we want to set max_iter to be the commensurate with the number of examples required\n",
+    "# for the model converge (as cited in the documentation)\n",
+    "\n",
+    "# it's a bit unclear how n_params relates to BOTH the number of data points required\n",
+    "# for the model to converge AND how many hyperparameters to try out (i.e. n_iter in RandomizedSearchCV)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import math"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 30,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Chunk size: 12500.0\n",
+      "Total chunks: 2\n",
+      "Last chunk size: 12500.0\n"
+     ]
+    }
+   ],
+   "source": [
+    "# choose chunk size so that the remainder is not a tiny number\n",
+    "print(f'Chunk size: {chunk_size}')\n",
+    "print(f'Total chunks: {math.ceil(len(X_train) / chunk_size)}')\n",
+    "last_chunk_size = len(X_train) % chunk_size\n",
+    "if last_chunk_size == 0: # i.e. chunk_size evenly divides X_train\n",
+    "    last_chunk_size = chunk_size\n",
+    "print(f'Last chunk size: {last_chunk_size}')\n",
+    "\n",
+    "assert (len(X_train) % chunk_size > 10 or len(X_train) % chunk_size == 0), 'Choose another chunk size'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import dask.array as da\n",
+    "\n",
+    "X = da.from_array(X_train, chunks=(chunk_size, X_train.shape[-1]))\n",
+    "y = da.from_array(y_train, chunks=(chunk_size))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<table>\n",
+       "<tr>\n",
+       "<td>\n",
+       "<table>\n",
+       "  <thead>\n",
+       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr><th> Bytes </th><td> 494.00 MB </td> <td> 247.00 MB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (25000, 2470) </td> <td> (12500, 2470) </td></tr>\n",
+       "    <tr><th> Count </th><td> 3 Tasks </td><td> 2 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</td>\n",
+       "<td>\n",
+       "<svg width=\"88\" height=\"170\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "\n",
+       "  <!-- Horizontal lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"38\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"60\" x2=\"38\" y2=\"60\" />\n",
+       "  <line x1=\"0\" y1=\"120\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Vertical lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"38\" y1=\"0\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Colored Rectangle -->\n",
+       "  <polygon points=\"0.000000,0.000000 38.529848,0.000000 38.529848,120.000000 0.000000,120.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "\n",
+       "  <!-- Text -->\n",
+       "  <text x=\"19.264924\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >2470</text>\n",
+       "  <text x=\"58.529848\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,58.529848,60.000000)\">25000</text>\n",
+       "</svg>\n",
+       "</td>\n",
+       "</tr>\n",
+       "</table>"
+      ],
+      "text/plain": [
+       "dask.array<array, shape=(25000, 2470), dtype=int64, chunksize=(12500, 2470), chunktype=numpy.ndarray>"
+      ]
+     },
+     "execution_count": 32,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "X"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch.nn as nn"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# reinitialize and set train_split=None to let hyperband handle validation set splitting\n",
+    "skorch_model = NeuralNetClassifier(\n",
+    "                CNN,\n",
+    "                device=device,\n",
+    "                lr=0.001,\n",
+    "                optimizer=optim.Adam,\n",
+    "                criterion=nn.NLLLoss,\n",
+    "                train_split=None, # let hyperband handle it\n",
+    "                module__n_filters=100,\n",
+    "                module__filter_sizes=(2, 3, 4),\n",
+    "                module__dropout=0.2,\n",
+    "                module__pretrained_embeddings=vocab.vectors,\n",
+    "                batch_size=32,\n",
+    "                verbose=2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 36,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from scipy.stats import loguniform"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# define parameter grid\n",
+    "params = {'module__filter_size': [(1,2,3), (2, 3, 4), (3, 4, 5)], \n",
+    "          'module__n_filters': [25,50,100],\n",
+    "          'module__dropout': loguniform(1e-1, 3e-1),\n",
+    "          'batch_size': [32, 64],\n",
+    "         }"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 38,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from dask_ml.model_selection import HyperbandSearchCV"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 39,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search = HyperbandSearchCV(\n",
+    "    skorch_model,\n",
+    "    params,\n",
+    "    max_iter=max_iter,\n",
+    "    verbose=True,\n",
+    "    test_size=0.2 # validation size\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 40,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "26"
+      ]
+     },
+     "execution_count": 40,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "search.metadata[\"partial_fit_calls\"]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 41,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import time"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# to clear up any confusion, every time partial_fit is called, we're passing in chunk_size number of\n",
+    "# data points. Then skorch handles the batch size either by being set explicitly or as part of the param grid.\n",
+    "\n",
+    "# to compare this grid search to number of epochs, we have 26 partial_fit calls * 10k data points = 260k examples\n",
+    "# with a training set size of 25k * .8 = 20k data points, this is 13 epochs!\n",
+    "# considering that it takes approximately 5 epochs to train a model, you would get through less than 3 sets of \n",
+    "# hyperparameters if manually searching. Instead we'll search through ~5."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[CV, bracket=1] creating 3 models\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 160.00 MB detected in task graph: \n",
+      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .1570]]),\n",
+      "), 0]\n",
+      "Consider scattering large objects ahead of time\n",
+      "with client.scatter to reduce scheduler burden and \n",
+      "keep data on workers\n",
+      "\n",
+      "    future = client.submit(func, big_data)    # bad\n",
+      "\n",
+      "    big_future = client.scatter(big_data)     # good\n",
+      "    future = client.submit(func, big_future)  # good\n",
+      "  % (format_bytes(len(b)), s)\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[CV, bracket=0] creating 2 models\n",
+      "[CV, bracket=0] For training there are between 10000 and 10000 examples in each chunk\n",
+      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "tornado.application - ERROR - Multiple exceptions in yield list\n",
+      "Traceback (most recent call last):\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 849, in callback\n",
+      "    result_list.append(f.result())\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
+      "    yielded = self.gen.throw(*exc_info)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 625, in _fit\n",
+      "    prefix=self.prefix,\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
+      "    value = future.result()\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
+      "    yielded = self.gen.throw(*exc_info)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 233, in _fit\n",
+      "    metas = yield client.gather(new_scores)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
+      "    value = future.result()\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/client.py\", line 1826, in _gather\n",
+      "    raise exception.with_traceback(traceback)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 116, in _create_model\n",
+      "    return model, {\"model_id\": ident, \"params\": params, \"partial_fit_calls\": 0}\n",
+      "  File \"/opt/conda/lib/python3.7/contextlib.py\", line 130, in __exit__\n",
+      "    self.gen.throw(type, value, traceback)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/utils.py\", line 676, in log_errors\n",
+      "    raise\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/utils.py\", line 676, in log_errors\n",
+      "    raise\n",
+      "  File \"/opt/conda/lib/python3.7/bdb.py\", line 88, in trace_dispatch\n",
+      "    return self.dispatch_line(frame)\n",
+      "  File \"/opt/conda/lib/python3.7/bdb.py\", line 113, in dispatch_line\n",
+      "    if self.quitting: raise BdbQuit\n",
+      "bdb.BdbQuit\n"
+     ]
+    },
+    {
+     "ename": "BdbQuit",
+     "evalue": "",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mBdbQuit\u001b[0m                                   Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-43-b1f07b9db6c0>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    671\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mdefault_client\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mif_delegate_has_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdelegate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_estimator_\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"estimator\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 347\u001b[0;31m         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    348\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    349\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mf\u001b[0;34m()\u001b[0m\n\u001b[1;32m    329\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mcallback_timeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    330\u001b[0m                 \u001b[0mfuture\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0masyncio\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait_for\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfuture\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 331\u001b[0;31m             \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    332\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    333\u001b[0m             \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_hyperband.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    398\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    399\u001b[0m         \u001b[0;31m# _fit is run in parallel because it's also a tornado coroutine\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 400\u001b[0;31m         \u001b[0m_SHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mSHAs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0m_brackets_ids\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    401\u001b[0m         \u001b[0mSHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_brackets_ids\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_SHAs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    402\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mcallback\u001b[0;34m(f)\u001b[0m\n\u001b[1;32m    847\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mchildren\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    848\u001b[0m                 \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 849\u001b[0;31m                     \u001b[0mresult_list\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    850\u001b[0m                 \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    851\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    623\u001b[0m             \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom_state\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    624\u001b[0m             \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 625\u001b[0;31m             \u001b[0mprefix\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprefix\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    626\u001b[0m         )\n\u001b[1;32m    627\u001b[0m         \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_process_results\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(model, params, X_train, y_train, X_test, y_test, additional_calls, fit_params, scorer, random_state, verbose, prefix)\u001b[0m\n\u001b[1;32m    231\u001b[0m     \u001b[0;31m# async for future, result in seq:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    232\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mitertools\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcount\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 233\u001b[0;31m         \u001b[0mmetas\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mclient\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnew_scores\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    235\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlog_delay\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlog_delay\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36m_gather\u001b[0;34m(self, futures, errors, direct, local_worker)\u001b[0m\n\u001b[1;32m   1824\u001b[0m                             \u001b[0mexc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCancelledError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1825\u001b[0m                         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1826\u001b[0;31m                             \u001b[0;32mraise\u001b[0m \u001b[0mexception\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1827\u001b[0m                         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1828\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0merrors\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m\"skip\"\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_create_model\u001b[0;34m()\u001b[0m\n\u001b[1;32m    114\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mlog_errors\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpdb\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    115\u001b[0m         \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mclone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mparams\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 116\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0;34m\"model_id\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mident\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"params\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mparams\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"partial_fit_calls\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    117\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    118\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/contextlib.py\u001b[0m in \u001b[0;36m__exit__\u001b[0;34m()\u001b[0m\n\u001b[1;32m    128\u001b[0m                 \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    129\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 130\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    131\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    132\u001b[0m                 \u001b[0;31m# Suppress StopIteration *unless* it's the same exception that\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mlog_errors\u001b[0;34m()\u001b[0m\n\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m             \u001b[0mpdb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_trace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 676\u001b[0;31m         \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mlog_errors\u001b[0;34m()\u001b[0m\n\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m             \u001b[0mpdb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_trace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 676\u001b[0;31m         \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/bdb.py\u001b[0m in \u001b[0;36mtrace_dispatch\u001b[0;34m()\u001b[0m\n\u001b[1;32m     86\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0;31m# None\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     87\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'line'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 88\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     89\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'call'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     90\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/bdb.py\u001b[0m in \u001b[0;36mdispatch_line\u001b[0;34m()\u001b[0m\n\u001b[1;32m    111\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstop_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbreak_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    112\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muser_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 113\u001b[0;31m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mquitting\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;32mraise\u001b[0m \u001b[0mBdbQuit\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    114\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrace_dispatch\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    115\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mBdbQuit\u001b[0m: "
+     ]
+    }
+   ],
+   "source": [
+    "# it's been erroring out with a less than helpful error message. This started happening right around \n",
+    "# when I started passing module__pretrained_embeddings=vocab.vectors. Unclear if this is the culprit.\n",
+    "\n",
+    "# NB: this took ~900 seconds on a single Nvidia GTX 980 Ti\n",
+    "# notice how the number of training datapoints relates to the chunk size and our test_size\n",
+    "# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\n",
+    "start = time.time()\n",
+    "search.fit(X, y)\n",
+    "end = time.time()\n",
+    "duration = round(end - start, 2)\n",
+    "print(f'Time to complete grid search: {duration} seconds')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Integration"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "`HyperbandSearchCV` follows the Scikit-learn API and mirrors Scikit-learn's `RandomizedSearchCV`. This means that it \"just works\". All the Scikit-learn attributes and methods are available:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.best_score_"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.best_estimator_"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.cv_results_"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "cv_results = pd.DataFrame(search.cv_results_)\n",
+    "cv_results.head()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.score(X_test, y_test)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.predict(X_test)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "search.predict(X_test).compute()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "It also has some other attributes."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "hist = pd.DataFrame(search.history_)\n",
+    "hist.head()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "This illustrates the history after every `partial_fit` call. There's also an attributed `model_history_` that records the history for each model (it's a reorganization of `history_`)."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Learn more"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "This notebook covered basic usage `HyperbandSearchCV`. The following documentation and resources might be useful to learn more about `HyperbandSearchCV`, including some of the finer use cases:\n",
+    "\n",
+    "* [A talk](https://www.youtube.com/watch?v=x67K9FiPFBQ) introducing `HyperbandSearchCV` to the SciPy 2019 audience and the [corresponding paper](https://conference.scipy.org/proceedings/scipy2019/pdfs/scott_sievert.pdf)\n",
+    "* [HyperbandSearchCV's documentation](https://ml.dask.org/modules/generated/dask_ml.model_selection.HyperbandSearchCV.html)\n",
+    "\n",
+    "Performance comparisons can be found in the SciPy 2019 talk/paper."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.7"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

From 171af05e0a53fd907ad536f3f8fc79e3621d7a4b Mon Sep 17 00:00:00 2001
From: Todd Morrill <tmorrill12@gmail.com>
Date: Sat, 23 May 2020 11:32:49 -0400
Subject: [PATCH 2/6] fix param grid typo and pretrained embeddings

---
 machine-learning/Untitled1.ipynb             |   6 -
 machine-learning/model.py                    |   4 +-
 machine-learning/skorch-hyperparam-opt.ipynb | 809 +++++++++++++++----
 3 files changed, 633 insertions(+), 186 deletions(-)
 delete mode 100644 machine-learning/Untitled1.ipynb

diff --git a/machine-learning/Untitled1.ipynb b/machine-learning/Untitled1.ipynb
deleted file mode 100644
index 2fd64429..00000000
--- a/machine-learning/Untitled1.ipynb
+++ /dev/null
@@ -1,6 +0,0 @@
-{
- "cells": [],
- "metadata": {},
- "nbformat": 4,
- "nbformat_minor": 2
-}
diff --git a/machine-learning/model.py b/machine-learning/model.py
index 54cc9192..55469aa5 100644
--- a/machine-learning/model.py
+++ b/machine-learning/model.py
@@ -2,13 +2,15 @@
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
+import torchtext
 
 
 class CNN(nn.Module):
     def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.2, pretrained_embeddings=None):
         
         super().__init__()
-        self.embedding = nn.Embedding.from_pretrained(torch.FloatTensor(pretrained_embeddings))
+        # will be used to initialize model embeddings layer
+        self.embedding = nn.Embedding.from_pretrained(pretrained_embeddings)
         self.embedding.weight.requires_grad = False # save some computation
         embedding_dim = self.embedding.embedding_dim
         self.conv_0 = nn.Conv1d(in_channels = 1, 
diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
index 54344473..d80e4fd3 100644
--- a/machine-learning/skorch-hyperparam-opt.ipynb
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -37,7 +37,7 @@
        "<td style=\"vertical-align: top; border: 0px solid white\">\n",
        "<h3 style=\"text-align: left;\">Client</h3>\n",
        "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
-       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42667</li>\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42413</li>\n",
        "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
        "</ul>\n",
        "</td>\n",
@@ -53,7 +53,7 @@
        "</table>"
       ],
       "text/plain": [
-       "<Client: 'tcp://127.0.0.1:42667' processes=1 threads=1, memory=33.68 GB>"
+       "<Client: 'tcp://127.0.0.1:42413' processes=1 threads=1, memory=33.68 GB>"
       ]
      },
      "execution_count": 2,
@@ -130,7 +130,7 @@
     "# takes approx. 10 minutes to download data and embeddings (will be cached for re-use)\n",
     "\n",
     "# set up fields\n",
-    "TEXT = data.Field(lower=True, include_lengths=True, batch_first=True)\n",
+    "TEXT = data.Field(lower=True, include_lengths=True, batch_first=True, )\n",
     "LABEL = data.Field(sequential=False, unk_token=None)\n",
     "\n",
     "# make splits for data\n",
@@ -146,7 +146,7 @@
     "\n",
     "# make iterator for splits\n",
     "train_iter, test_iter = data.BucketIterator.splits(\n",
-    "    (train, test), batch_sizes=(32, 64), device=device)"
+    "    (train, test), batch_sizes=(32, 64), device='cpu')"
    ]
   },
   {
@@ -222,16 +222,16 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "tensor([[  45,  123,   25,  ...,    1,    1,    1],\n",
-      "        [5311, 3868,    2,  ...,    1,    1,    1],\n",
-      "        [ 133,    8,   42,  ...,    1,    1,    1],\n",
+      "tensor([[  542,  5101,    41,  ...,     1,     1,     1],\n",
+      "        [   10,     7,    31,  ...,     1,     1,     1],\n",
+      "        [    0,    49, 15842,  ...,     1,     1,     1],\n",
       "        ...,\n",
-      "        [   9,  176,   10,  ...,    1,    1,    1],\n",
-      "        [1467,   10,   30,  ...,    1,    1,    1],\n",
-      "        [  82, 1297,    9,  ...,    1,    1,    1]], device='cuda:0')\n",
-      "tensor([147, 347, 361,  66, 147, 262, 336, 105, 255, 132, 128, 162, 122, 184,\n",
-      "        195, 100, 226, 127, 152, 159, 120, 204, 267, 209, 129, 486,  95, 151,\n",
-      "        140, 247, 127, 167], device='cuda:0')\n"
+      "        [   25,    86,   311,  ...,     1,     1,     1],\n",
+      "        [    9,   157,   168,  ...,     1,     1,     1],\n",
+      "        [   10,    49,   235,  ...,     1,     1,     1]])\n",
+      "tensor([132, 121,  72, 343, 112,  90, 226, 982, 118, 341, 145,  64, 149,  87,\n",
+      "        244, 270, 122,  94, 141,  76, 154, 151, 429,  97, 104,  76, 120, 125,\n",
+      "        120, 118, 134, 154])\n"
      ]
     }
    ],
@@ -250,8 +250,8 @@
     {
      "data": {
       "text/plain": [
-       "tensor([1, 0, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0,\n",
-       "        1, 1, 1, 1, 0, 1, 0, 0], device='cuda:0')"
+       "tensor([1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0,\n",
+       "        0, 1, 0, 1, 1, 0, 1, 1])"
       ]
      },
      "execution_count": 12,
@@ -311,38 +311,38 @@
     {
      "data": {
       "text/plain": [
-       "tensor([[0.4709, 0.5291],\n",
-       "        [0.2993, 0.7007],\n",
-       "        [0.2609, 0.7391],\n",
-       "        [0.3476, 0.6524],\n",
-       "        [0.4323, 0.5677],\n",
-       "        [0.3394, 0.6606],\n",
-       "        [0.2745, 0.7255],\n",
-       "        [0.3152, 0.6848],\n",
-       "        [0.4568, 0.5432],\n",
-       "        [0.4189, 0.5811],\n",
-       "        [0.4351, 0.5649],\n",
-       "        [0.4204, 0.5796],\n",
-       "        [0.3963, 0.6037],\n",
-       "        [0.3482, 0.6518],\n",
-       "        [0.3528, 0.6472],\n",
-       "        [0.4137, 0.5863],\n",
-       "        [0.3327, 0.6673],\n",
-       "        [0.3380, 0.6620],\n",
-       "        [0.4874, 0.5126],\n",
-       "        [0.2904, 0.7096],\n",
-       "        [0.2868, 0.7132],\n",
-       "        [0.3196, 0.6804],\n",
-       "        [0.3689, 0.6311],\n",
-       "        [0.2525, 0.7475],\n",
-       "        [0.2493, 0.7507],\n",
-       "        [0.3308, 0.6692],\n",
-       "        [0.2088, 0.7912],\n",
-       "        [0.4834, 0.5166],\n",
-       "        [0.2363, 0.7637],\n",
-       "        [0.4511, 0.5489],\n",
-       "        [0.2282, 0.7718],\n",
-       "        [0.3502, 0.6498]], device='cuda:0', grad_fn=<SoftmaxBackward>)"
+       "tensor([[0.4698, 0.5302],\n",
+       "        [0.3791, 0.6209],\n",
+       "        [0.2715, 0.7285],\n",
+       "        [0.3220, 0.6780],\n",
+       "        [0.3979, 0.6021],\n",
+       "        [0.4197, 0.5803],\n",
+       "        [0.3409, 0.6591],\n",
+       "        [0.2695, 0.7305],\n",
+       "        [0.4400, 0.5600],\n",
+       "        [0.4209, 0.5791],\n",
+       "        [0.4640, 0.5360],\n",
+       "        [0.4475, 0.5525],\n",
+       "        [0.4261, 0.5739],\n",
+       "        [0.3032, 0.6968],\n",
+       "        [0.3452, 0.6548],\n",
+       "        [0.3803, 0.6197],\n",
+       "        [0.3741, 0.6259],\n",
+       "        [0.3796, 0.6204],\n",
+       "        [0.5005, 0.4995],\n",
+       "        [0.3706, 0.6294],\n",
+       "        [0.3142, 0.6858],\n",
+       "        [0.3772, 0.6228],\n",
+       "        [0.3281, 0.6719],\n",
+       "        [0.3190, 0.6810],\n",
+       "        [0.3340, 0.6660],\n",
+       "        [0.3948, 0.6052],\n",
+       "        [0.2351, 0.7649],\n",
+       "        [0.4538, 0.5462],\n",
+       "        [0.2321, 0.7679],\n",
+       "        [0.4226, 0.5774],\n",
+       "        [0.2449, 0.7551],\n",
+       "        [0.3471, 0.6529]], device='cuda:0', grad_fn=<SoftmaxBackward>)"
       ]
      },
      "execution_count": 15,
@@ -352,8 +352,9 @@
    ],
    "source": [
     "# smoketest\n",
-    "model = CNN(pretrained_embeddings=vocab.vectors).to(device)\n",
-    "model(batch.text[0].to(device))"
+    "model = CNN(pretrained_embeddings=TEXT.vocab.vectors).to(device)\n",
+    "gpu_batch = batch.text[0].to(device)\n",
+    "model(gpu_batch)"
    ]
   },
   {
@@ -365,6 +366,24 @@
     "del model"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "del gpu_batch"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "torch.cuda.empty_cache()"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -374,7 +393,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 19,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -385,7 +404,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
+   "execution_count": 20,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -404,7 +423,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -423,7 +442,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [
     {
@@ -432,7 +451,7 @@
        "(25000, 2470)"
       ]
      },
-     "execution_count": 20,
+     "execution_count": 22,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -444,7 +463,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -453,7 +472,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -470,14 +489,14 @@
     "                module__n_filters=100,\n",
     "                module__filter_sizes=(2,3,4),\n",
     "                module__dropout=0.2,\n",
-    "                module__pretrained_embeddings=vocab.vectors,\n",
+    "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
     "                batch_size=32,\n",
     "                verbose=2)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [
     {
@@ -486,8 +505,8 @@
      "text": [
       "  epoch    train_loss    valid_acc    valid_loss      dur\n",
       "-------  ------------  -----------  ------------  -------\n",
-      "      1        \u001b[36m0.6302\u001b[0m       \u001b[32m0.7516\u001b[0m        \u001b[35m0.5243\u001b[0m  66.5468\n",
-      "      2        \u001b[36m0.4797\u001b[0m       \u001b[32m0.7994\u001b[0m        \u001b[35m0.4413\u001b[0m  59.9177\n"
+      "      1        \u001b[36m0.4689\u001b[0m       \u001b[32m0.8370\u001b[0m        \u001b[35m0.3656\u001b[0m  60.5744\n",
+      "      2        \u001b[36m0.3466\u001b[0m       \u001b[32m0.8524\u001b[0m        \u001b[35m0.3413\u001b[0m  58.6526\n"
      ]
     },
     {
@@ -495,7 +514,7 @@
       "text/plain": [
        "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
        "  module_=CNN(\n",
-       "    (embedding): Embedding(400000, 100)\n",
+       "    (embedding): Embedding(25002, 100)\n",
        "    (conv_0): Conv1d(1, 100, kernel_size=(2, 100), stride=(1,))\n",
        "    (conv_1): Conv1d(1, 100, kernel_size=(3, 100), stride=(1,))\n",
        "    (conv_2): Conv1d(1, 100, kernel_size=(4, 100), stride=(1,))\n",
@@ -505,7 +524,7 @@
        ")"
       ]
      },
-     "execution_count": 23,
+     "execution_count": 25,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -516,16 +535,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 26,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "0.79112"
+       "0.85512"
       ]
      },
-     "execution_count": 24,
+     "execution_count": 26,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -537,7 +556,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": 27,
    "metadata": {},
    "outputs": [
     {
@@ -546,7 +565,7 @@
        "(array([0, 1]), array([12500, 12500]))"
       ]
      },
-     "execution_count": 25,
+     "execution_count": 27,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -558,10 +577,13 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 28,
    "metadata": {},
    "outputs": [],
    "source": [
+    "# NB: this has no effect on GPU memory usage. If I keyboard interrupt, the workers get\n",
+    "# restarted and memory usage goes down. Deleting these \"handler\" objects doesn't delete\n",
+    "# GPU memory references on the workers. \n",
     "del skorch_model"
    ]
   },
@@ -574,7 +596,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": 29,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -591,7 +613,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 28,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -604,7 +626,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -613,7 +635,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 32,
    "metadata": {},
    "outputs": [
     {
@@ -640,7 +662,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 33,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -652,7 +674,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 34,
    "metadata": {},
    "outputs": [
     {
@@ -700,7 +722,7 @@
        "dask.array<array, shape=(25000, 2470), dtype=int64, chunksize=(12500, 2470), chunktype=numpy.ndarray>"
       ]
      },
-     "execution_count": 32,
+     "execution_count": 34,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -709,15 +731,6 @@
     "X"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": 33,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import torch.nn as nn"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": 35,
@@ -735,7 +748,7 @@
     "                module__n_filters=100,\n",
     "                module__filter_sizes=(2, 3, 4),\n",
     "                module__dropout=0.2,\n",
-    "                module__pretrained_embeddings=vocab.vectors,\n",
+    "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
     "                batch_size=32,\n",
     "                verbose=2)"
    ]
@@ -756,8 +769,8 @@
    "outputs": [],
    "source": [
     "# define parameter grid\n",
-    "params = {'module__filter_size': [(1,2,3), (2, 3, 4), (3, 4, 5)], \n",
-    "          'module__n_filters': [25,50,100],\n",
+    "params = {'module__filter_sizes': [(1, 2, 3), (2, 3, 4), (3, 4, 5)], \n",
+    "          'module__n_filters': [25, 50, 100],\n",
     "          'module__dropout': loguniform(1e-1, 3e-1),\n",
     "          'batch_size': [32, 64],\n",
     "         }"
@@ -811,6 +824,26 @@
    "cell_type": "code",
    "execution_count": 41,
    "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "5"
+      ]
+     },
+     "execution_count": 41,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "search.metadata['n_models']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
    "outputs": [],
    "source": [
     "import time"
@@ -818,7 +851,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 42,
+   "execution_count": 43,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -833,7 +866,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 43,
+   "execution_count": 44,
    "metadata": {},
    "outputs": [
     {
@@ -847,8 +880,8 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 160.00 MB detected in task graph: \n",
-      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .1570]]),\n",
+      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 10.00 MB detected in task graph: \n",
+      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .0721]]),\n",
       "), 0]\n",
       "Consider scattering large objects ahead of time\n",
       "with client.scatter to reduce scheduler burden and \n",
@@ -867,77 +900,13 @@
      "text": [
       "[CV, bracket=0] creating 2 models\n",
       "[CV, bracket=0] For training there are between 10000 and 10000 examples in each chunk\n",
-      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "tornado.application - ERROR - Multiple exceptions in yield list\n",
-      "Traceback (most recent call last):\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 849, in callback\n",
-      "    result_list.append(f.result())\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
-      "    yielded = self.gen.throw(*exc_info)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 625, in _fit\n",
-      "    prefix=self.prefix,\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
-      "    value = future.result()\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
-      "    yielded = self.gen.throw(*exc_info)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 233, in _fit\n",
-      "    metas = yield client.gather(new_scores)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
-      "    value = future.result()\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/client.py\", line 1826, in _gather\n",
-      "    raise exception.with_traceback(traceback)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 116, in _create_model\n",
-      "    return model, {\"model_id\": ident, \"params\": params, \"partial_fit_calls\": 0}\n",
-      "  File \"/opt/conda/lib/python3.7/contextlib.py\", line 130, in __exit__\n",
-      "    self.gen.throw(type, value, traceback)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/utils.py\", line 676, in log_errors\n",
-      "    raise\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/utils.py\", line 676, in log_errors\n",
-      "    raise\n",
-      "  File \"/opt/conda/lib/python3.7/bdb.py\", line 88, in trace_dispatch\n",
-      "    return self.dispatch_line(frame)\n",
-      "  File \"/opt/conda/lib/python3.7/bdb.py\", line 113, in dispatch_line\n",
-      "    if self.quitting: raise BdbQuit\n",
-      "bdb.BdbQuit\n"
-     ]
-    },
-    {
-     "ename": "BdbQuit",
-     "evalue": "",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mBdbQuit\u001b[0m                                   Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-43-b1f07b9db6c0>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    671\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mdefault_client\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mif_delegate_has_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdelegate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_estimator_\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"estimator\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 347\u001b[0;31m         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    348\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    349\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mf\u001b[0;34m()\u001b[0m\n\u001b[1;32m    329\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mcallback_timeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    330\u001b[0m                 \u001b[0mfuture\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0masyncio\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait_for\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfuture\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 331\u001b[0;31m             \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    332\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    333\u001b[0m             \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_hyperband.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    398\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    399\u001b[0m         \u001b[0;31m# _fit is run in parallel because it's also a tornado coroutine\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 400\u001b[0;31m         \u001b[0m_SHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mSHAs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0m_brackets_ids\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    401\u001b[0m         \u001b[0mSHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_brackets_ids\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_SHAs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    402\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mcallback\u001b[0;34m(f)\u001b[0m\n\u001b[1;32m    847\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mchildren\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    848\u001b[0m                 \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 849\u001b[0;31m                     \u001b[0mresult_list\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    850\u001b[0m                 \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    851\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    623\u001b[0m             \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom_state\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    624\u001b[0m             \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 625\u001b[0;31m             \u001b[0mprefix\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprefix\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    626\u001b[0m         )\n\u001b[1;32m    627\u001b[0m         \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_process_results\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(model, params, X_train, y_train, X_test, y_test, additional_calls, fit_params, scorer, random_state, verbose, prefix)\u001b[0m\n\u001b[1;32m    231\u001b[0m     \u001b[0;31m# async for future, result in seq:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    232\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mitertools\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcount\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 233\u001b[0;31m         \u001b[0mmetas\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mclient\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnew_scores\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    235\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlog_delay\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlog_delay\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36m_gather\u001b[0;34m(self, futures, errors, direct, local_worker)\u001b[0m\n\u001b[1;32m   1824\u001b[0m                             \u001b[0mexc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCancelledError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1825\u001b[0m                         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1826\u001b[0;31m                             \u001b[0;32mraise\u001b[0m \u001b[0mexception\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1827\u001b[0m                         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1828\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0merrors\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m\"skip\"\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_create_model\u001b[0;34m()\u001b[0m\n\u001b[1;32m    114\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mlog_errors\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpdb\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    115\u001b[0m         \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mclone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mparams\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 116\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0;34m\"model_id\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mident\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"params\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mparams\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"partial_fit_calls\"\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    117\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    118\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/contextlib.py\u001b[0m in \u001b[0;36m__exit__\u001b[0;34m()\u001b[0m\n\u001b[1;32m    128\u001b[0m                 \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    129\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 130\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    131\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    132\u001b[0m                 \u001b[0;31m# Suppress StopIteration *unless* it's the same exception that\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mlog_errors\u001b[0;34m()\u001b[0m\n\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m             \u001b[0mpdb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_trace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 676\u001b[0;31m         \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mlog_errors\u001b[0;34m()\u001b[0m\n\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m             \u001b[0mpdb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_trace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 676\u001b[0;31m         \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/bdb.py\u001b[0m in \u001b[0;36mtrace_dispatch\u001b[0;34m()\u001b[0m\n\u001b[1;32m     86\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0;31m# None\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     87\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'line'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 88\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     89\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'call'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     90\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/bdb.py\u001b[0m in \u001b[0;36mdispatch_line\u001b[0;34m()\u001b[0m\n\u001b[1;32m    111\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstop_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbreak_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    112\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muser_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 113\u001b[0;31m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mquitting\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;32mraise\u001b[0m \u001b[0mBdbQuit\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    114\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrace_dispatch\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    115\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;31mBdbQuit\u001b[0m: "
+      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n",
+      "[CV, bracket=0] validation score of 0.8386 received after 1 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8306 received after 1 partial_fit calls\n",
+      "[CV, bracket=0] validation score of 0.8446 received after 8 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8430 received after 2 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8456 received after 6 partial_fit calls\n",
+      "Time to complete grid search: 3813.07 seconds\n"
      ]
     }
    ],
@@ -945,7 +914,7 @@
     "# it's been erroring out with a less than helpful error message. This started happening right around \n",
     "# when I started passing module__pretrained_embeddings=vocab.vectors. Unclear if this is the culprit.\n",
     "\n",
-    "# NB: this took ~900 seconds on a single Nvidia GTX 980 Ti\n",
+    "# NB: this took ~3800 seconds on a single Nvidia GTX 980 Ti\n",
     "# notice how the number of training datapoints relates to the chunk size and our test_size\n",
     "# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\n",
     "start = time.time()\n",
@@ -955,6 +924,26 @@
     "print(f'Time to complete grid search: {duration} seconds')"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 50,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "63.333333333333336"
+      ]
+     },
+     "execution_count": 50,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "3800 / 60"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -971,25 +960,56 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 45,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "0.8456"
+      ]
+     },
+     "execution_count": 45,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.best_score_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 46,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
+       "  module_=CNN(\n",
+       "    (embedding): Embedding(25002, 100)\n",
+       "    (conv_0): Conv1d(1, 50, kernel_size=(3, 100), stride=(1,))\n",
+       "    (conv_1): Conv1d(1, 50, kernel_size=(4, 100), stride=(1,))\n",
+       "    (conv_2): Conv1d(1, 50, kernel_size=(5, 100), stride=(1,))\n",
+       "    (fc): Linear(in_features=150, out_features=2, bias=True)\n",
+       "    (dropout): Dropout(p=0.22181138802671369, inplace=False)\n",
+       "  ),\n",
+       ")"
+      ]
+     },
+     "execution_count": 46,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.best_estimator_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 47,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -998,18 +1018,70 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 48,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'param_module__n_filters': array([50, 50, 50, 25, 50]),\n",
+       " 'test_score': array([0.8456, 0.7982, 0.733 , 0.843 , 0.8446]),\n",
+       " 'std_score_time': array([0.06216696, 0.19731486, 0.56393194, 1.33936262, 0.06445384]),\n",
+       " 'rank_test_score': array([1, 2, 3, 2, 1]),\n",
+       " 'mean_score_time': array([3.82462454, 4.0282222 , 5.47523713, 4.41082978, 5.29884744]),\n",
+       " 'params': array([{'batch_size': 64, 'module__dropout': 0.22181138802671369, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.19188105885296006, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.11893575283135104, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.10900910968190201, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 25},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.10009422737462631, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50}],\n",
+       "       dtype=object),\n",
+       " 'partial_fit_calls': array([6, 2, 2, 8, 8]),\n",
+       " 'param_module__dropout': array([0.22181139, 0.19188106, 0.11893575, 0.10900911, 0.10009423]),\n",
+       " 'std_partial_fit_time': array([ 3.6288025 ,  1.52035153, 11.09951091, 15.07569551,  5.98224938]),\n",
+       " 'mean_partial_fit_time': array([ 93.61417103,  96.36260712, 136.89576507, 143.69626164,\n",
+       "        130.64727747]),\n",
+       " 'model_id': array(['bracket=1-0', 'bracket=1-1', 'bracket=1-2', 'bracket=0-0',\n",
+       "        'bracket=0-1'], dtype='<U11'),\n",
+       " 'bracket': array([1, 1, 1, 0, 0]),\n",
+       " 'param_batch_size': array([64, 64, 32, 32, 32]),\n",
+       " 'param_module__filter_sizes': array([[3, 4, 5],\n",
+       "        [3, 4, 5],\n",
+       "        [3, 4, 5],\n",
+       "        [2, 3, 4],\n",
+       "        [3, 4, 5]])}"
+      ]
+     },
+     "execution_count": 48,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.cv_results_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 49,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "ename": "Exception",
+     "evalue": "Data must be 1-dimensional",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mException\u001b[0m                                 Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-49-245daea12296>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mcv_results\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDataFrame\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcv_results_\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mcv_results\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhead\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/frame.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, data, index, columns, dtype, copy)\u001b[0m\n\u001b[1;32m    409\u001b[0m             )\n\u001b[1;32m    410\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 411\u001b[0;31m             \u001b[0mmgr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0minit_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    412\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mma\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mMaskedArray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    413\u001b[0m             \u001b[0;32mimport\u001b[0m \u001b[0mnumpy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mma\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmrecords\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mmrecords\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36minit_dict\u001b[0;34m(data, index, columns, dtype)\u001b[0m\n\u001b[1;32m    255\u001b[0m             \u001b[0marr\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mis_datetime64tz_dtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marr\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0marr\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marrays\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    256\u001b[0m         ]\n\u001b[0;32m--> 257\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0marrays_to_mgr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marrays\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdata_names\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    258\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    259\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36marrays_to_mgr\u001b[0;34m(arrays, arr_names, index, columns, dtype)\u001b[0m\n\u001b[1;32m     80\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     81\u001b[0m     \u001b[0;31m# don't force copy because getting jammed in an ndarray anyway\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 82\u001b[0;31m     \u001b[0marrays\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_homogenize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marrays\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     83\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     84\u001b[0m     \u001b[0;31m# from BlockManager perspective\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36m_homogenize\u001b[0;34m(data, index, dtype)\u001b[0m\n\u001b[1;32m    321\u001b[0m                 \u001b[0mval\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlib\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfast_multiget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moindex\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdefault\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnan\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    322\u001b[0m             val = sanitize_array(\n\u001b[0;32m--> 323\u001b[0;31m                 \u001b[0mval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mraise_cast_failure\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    324\u001b[0m             )\n\u001b[1;32m    325\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36msanitize_array\u001b[0;34m(data, index, dtype, copy, raise_cast_failure)\u001b[0m\n\u001b[1;32m    727\u001b[0m     \u001b[0;32melif\u001b[0m \u001b[0msubarr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    728\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndarray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 729\u001b[0;31m             \u001b[0;32mraise\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Data must be 1-dimensional\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    730\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    731\u001b[0m             \u001b[0msubarr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masarray_tuplesafe\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mException\u001b[0m: Data must be 1-dimensional"
+     ]
+    }
+   ],
    "source": [
     "cv_results = pd.DataFrame(search.cv_results_)\n",
     "cv_results.head()"
@@ -1017,27 +1089,287 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 60,
    "metadata": {},
    "outputs": [],
+   "source": [
+    "# issue with numpy converting array of tuples into a 2d array\n",
+    "search.cv_results_['param_module__filter_sizes'] = search.cv_results_['param_module__filter_sizes'].tolist()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 61,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>param_module__n_filters</th>\n",
+       "      <th>test_score</th>\n",
+       "      <th>std_score_time</th>\n",
+       "      <th>rank_test_score</th>\n",
+       "      <th>mean_score_time</th>\n",
+       "      <th>params</th>\n",
+       "      <th>partial_fit_calls</th>\n",
+       "      <th>param_module__dropout</th>\n",
+       "      <th>std_partial_fit_time</th>\n",
+       "      <th>mean_partial_fit_time</th>\n",
+       "      <th>model_id</th>\n",
+       "      <th>bracket</th>\n",
+       "      <th>param_batch_size</th>\n",
+       "      <th>param_module__filter_sizes</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>50</td>\n",
+       "      <td>0.8456</td>\n",
+       "      <td>0.062167</td>\n",
+       "      <td>1</td>\n",
+       "      <td>3.824625</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.221811...</td>\n",
+       "      <td>6</td>\n",
+       "      <td>0.221811</td>\n",
+       "      <td>3.628802</td>\n",
+       "      <td>93.614171</td>\n",
+       "      <td>bracket=1-0</td>\n",
+       "      <td>1</td>\n",
+       "      <td>64</td>\n",
+       "      <td>[3, 4, 5]</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>50</td>\n",
+       "      <td>0.7982</td>\n",
+       "      <td>0.197315</td>\n",
+       "      <td>2</td>\n",
+       "      <td>4.028222</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.191881...</td>\n",
+       "      <td>2</td>\n",
+       "      <td>0.191881</td>\n",
+       "      <td>1.520352</td>\n",
+       "      <td>96.362607</td>\n",
+       "      <td>bracket=1-1</td>\n",
+       "      <td>1</td>\n",
+       "      <td>64</td>\n",
+       "      <td>[3, 4, 5]</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>50</td>\n",
+       "      <td>0.7330</td>\n",
+       "      <td>0.563932</td>\n",
+       "      <td>3</td>\n",
+       "      <td>5.475237</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.118935...</td>\n",
+       "      <td>2</td>\n",
+       "      <td>0.118936</td>\n",
+       "      <td>11.099511</td>\n",
+       "      <td>136.895765</td>\n",
+       "      <td>bracket=1-2</td>\n",
+       "      <td>1</td>\n",
+       "      <td>32</td>\n",
+       "      <td>[3, 4, 5]</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>25</td>\n",
+       "      <td>0.8430</td>\n",
+       "      <td>1.339363</td>\n",
+       "      <td>2</td>\n",
+       "      <td>4.410830</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.109009...</td>\n",
+       "      <td>8</td>\n",
+       "      <td>0.109009</td>\n",
+       "      <td>15.075696</td>\n",
+       "      <td>143.696262</td>\n",
+       "      <td>bracket=0-0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>32</td>\n",
+       "      <td>[2, 3, 4]</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>50</td>\n",
+       "      <td>0.8446</td>\n",
+       "      <td>0.064454</td>\n",
+       "      <td>1</td>\n",
+       "      <td>5.298847</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.100094...</td>\n",
+       "      <td>8</td>\n",
+       "      <td>0.100094</td>\n",
+       "      <td>5.982249</td>\n",
+       "      <td>130.647277</td>\n",
+       "      <td>bracket=0-1</td>\n",
+       "      <td>0</td>\n",
+       "      <td>32</td>\n",
+       "      <td>[3, 4, 5]</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   param_module__n_filters  test_score  std_score_time  rank_test_score  \\\n",
+       "0                       50      0.8456        0.062167                1   \n",
+       "1                       50      0.7982        0.197315                2   \n",
+       "2                       50      0.7330        0.563932                3   \n",
+       "3                       25      0.8430        1.339363                2   \n",
+       "4                       50      0.8446        0.064454                1   \n",
+       "\n",
+       "   mean_score_time                                             params  \\\n",
+       "0         3.824625  {'batch_size': 64, 'module__dropout': 0.221811...   \n",
+       "1         4.028222  {'batch_size': 64, 'module__dropout': 0.191881...   \n",
+       "2         5.475237  {'batch_size': 32, 'module__dropout': 0.118935...   \n",
+       "3         4.410830  {'batch_size': 32, 'module__dropout': 0.109009...   \n",
+       "4         5.298847  {'batch_size': 32, 'module__dropout': 0.100094...   \n",
+       "\n",
+       "   partial_fit_calls  param_module__dropout  std_partial_fit_time  \\\n",
+       "0                  6               0.221811              3.628802   \n",
+       "1                  2               0.191881              1.520352   \n",
+       "2                  2               0.118936             11.099511   \n",
+       "3                  8               0.109009             15.075696   \n",
+       "4                  8               0.100094              5.982249   \n",
+       "\n",
+       "   mean_partial_fit_time     model_id  bracket  param_batch_size  \\\n",
+       "0              93.614171  bracket=1-0        1                64   \n",
+       "1              96.362607  bracket=1-1        1                64   \n",
+       "2             136.895765  bracket=1-2        1                32   \n",
+       "3             143.696262  bracket=0-0        0                32   \n",
+       "4             130.647277  bracket=0-1        0                32   \n",
+       "\n",
+       "  param_module__filter_sizes  \n",
+       "0                  [3, 4, 5]  \n",
+       "1                  [3, 4, 5]  \n",
+       "2                  [3, 4, 5]  \n",
+       "3                  [2, 3, 4]  \n",
+       "4                  [3, 4, 5]  "
+      ]
+     },
+     "execution_count": 61,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "cv_results = pd.DataFrame(search.cv_results_)\n",
+    "cv_results.head()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 62,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "0.83816"
+      ]
+     },
+     "execution_count": 62,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.score(X_test, y_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 63,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<table>\n",
+       "<tr>\n",
+       "<td>\n",
+       "<table>\n",
+       "  <thead>\n",
+       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr><th> Bytes </th><td> 200.00 kB </td> <td> 200.00 kB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (25000,) </td> <td> (25000,) </td></tr>\n",
+       "    <tr><th> Count </th><td> 2 Tasks </td><td> 1 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</td>\n",
+       "<td>\n",
+       "<svg width=\"170\" height=\"75\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "\n",
+       "  <!-- Horizontal lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"120\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"25\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Vertical lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"120\" y1=\"0\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Colored Rectangle -->\n",
+       "  <polygon points=\"0.000000,0.000000 120.000000,0.000000 120.000000,25.412617 0.000000,25.412617\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "\n",
+       "  <!-- Text -->\n",
+       "  <text x=\"60.000000\" y=\"45.412617\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >25000</text>\n",
+       "  <text x=\"140.000000\" y=\"12.706308\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(0,140.000000,12.706308)\">1</text>\n",
+       "</svg>\n",
+       "</td>\n",
+       "</tr>\n",
+       "</table>"
+      ],
+      "text/plain": [
+       "dask.array<_predict, shape=(25000,), dtype=int64, chunksize=(25000,), chunktype=numpy.ndarray>"
+      ]
+     },
+     "execution_count": 63,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.predict(X_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 64,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([0, 1, 1, ..., 0, 0, 0])"
+      ]
+     },
+     "execution_count": 64,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.predict(X_test).compute()"
    ]
@@ -1051,9 +1383,128 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 65,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>model_id</th>\n",
+       "      <th>params</th>\n",
+       "      <th>partial_fit_calls</th>\n",
+       "      <th>partial_fit_time</th>\n",
+       "      <th>score</th>\n",
+       "      <th>score_time</th>\n",
+       "      <th>elapsed_wall_time</th>\n",
+       "      <th>bracket</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>bracket=0-0</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.109009...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>128.620566</td>\n",
+       "      <td>0.8272</td>\n",
+       "      <td>3.071467</td>\n",
+       "      <td>358.527012</td>\n",
+       "      <td>0</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>bracket=0-1</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.100094...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>124.665028</td>\n",
+       "      <td>0.8386</td>\n",
+       "      <td>5.234394</td>\n",
+       "      <td>358.527018</td>\n",
+       "      <td>0</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>bracket=1-0</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.221811...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>95.602639</td>\n",
+       "      <td>0.7558</td>\n",
+       "      <td>3.739039</td>\n",
+       "      <td>460.180206</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>bracket=1-1</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.191881...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>94.842256</td>\n",
+       "      <td>0.7496</td>\n",
+       "      <td>4.225537</td>\n",
+       "      <td>460.180212</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>bracket=1-2</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.118935...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>125.796254</td>\n",
+       "      <td>0.8306</td>\n",
+       "      <td>6.039169</td>\n",
+       "      <td>460.180215</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "      model_id                                             params  \\\n",
+       "0  bracket=0-0  {'batch_size': 32, 'module__dropout': 0.109009...   \n",
+       "1  bracket=0-1  {'batch_size': 32, 'module__dropout': 0.100094...   \n",
+       "2  bracket=1-0  {'batch_size': 64, 'module__dropout': 0.221811...   \n",
+       "3  bracket=1-1  {'batch_size': 64, 'module__dropout': 0.191881...   \n",
+       "4  bracket=1-2  {'batch_size': 32, 'module__dropout': 0.118935...   \n",
+       "\n",
+       "   partial_fit_calls  partial_fit_time   score  score_time  elapsed_wall_time  \\\n",
+       "0                  1        128.620566  0.8272    3.071467         358.527012   \n",
+       "1                  1        124.665028  0.8386    5.234394         358.527018   \n",
+       "2                  1         95.602639  0.7558    3.739039         460.180206   \n",
+       "3                  1         94.842256  0.7496    4.225537         460.180212   \n",
+       "4                  1        125.796254  0.8306    6.039169         460.180215   \n",
+       "\n",
+       "   bracket  \n",
+       "0        0  \n",
+       "1        0  \n",
+       "2        1  \n",
+       "3        1  \n",
+       "4        1  "
+      ]
+     },
+     "execution_count": 65,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "hist = pd.DataFrame(search.history_)\n",
     "hist.head()"

From 1263e2c1ab759f6301488c4293eefb3830fa6095 Mon Sep 17 00:00:00 2001
From: Todd Morrill <tmorrill12@gmail.com>
Date: Sat, 23 May 2020 21:54:07 -0400
Subject: [PATCH 3/6] implement custom DataLoader to reduce Skorch run time

---
 machine-learning/model.py                    |   7 +-
 machine-learning/skorch-hyperparam-opt.ipynb | 374 ++++++++++++++-----
 2 files changed, 292 insertions(+), 89 deletions(-)

diff --git a/machine-learning/model.py b/machine-learning/model.py
index 55469aa5..7ad44ea4 100644
--- a/machine-learning/model.py
+++ b/machine-learning/model.py
@@ -2,6 +2,7 @@
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
+from torch.nn.utils.rnn import pad_sequence
 import torchtext
 
 
@@ -26,8 +27,10 @@ def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.
         self.dropout = nn.Dropout(dropout)
         
     def forward(self, text):
+        # padding_value ought to be 1: TEXT.vocab.stoi[TEXT.pad_token]
+        padded_batch = pad_sequence(text, batch_first=True, padding_value=1)
         #text = [batch size, sent len]
-        embedded = self.embedding(text)
+        embedded = self.embedding(padded_batch)
         #embedded = [batch size, sent len, emb dim]
         embedded = embedded.unsqueeze(1)
         #embedded = [batch size, 1, sent len, emb dim]
@@ -43,4 +46,4 @@ def forward(self, text):
         #cat = [batch size, n_filters * len(filter_sizes)]
         logits = self.fc(cat)
         #logits = [batch_size, output_dim]
-        return F.softmax(logits, dim=-1)
\ No newline at end of file
+        return torch.log(F.softmax(logits, dim=-1))
\ No newline at end of file
diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
index d80e4fd3..69f6d639 100644
--- a/machine-learning/skorch-hyperparam-opt.ipynb
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -26,7 +26,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
@@ -37,7 +37,7 @@
        "<td style=\"vertical-align: top; border: 0px solid white\">\n",
        "<h3 style=\"text-align: left;\">Client</h3>\n",
        "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
-       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42413</li>\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:41949</li>\n",
        "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
        "</ul>\n",
        "</td>\n",
@@ -53,10 +53,10 @@
        "</table>"
       ],
       "text/plain": [
-       "<Client: 'tcp://127.0.0.1:42413' processes=1 threads=1, memory=33.68 GB>"
+       "<Client: 'tcp://127.0.0.1:41949' processes=1 threads=1, memory=33.68 GB>"
       ]
      },
-     "execution_count": 2,
+     "execution_count": 1,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -78,7 +78,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -103,7 +103,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -114,7 +114,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -123,14 +123,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
     "# takes approx. 10 minutes to download data and embeddings (will be cached for re-use)\n",
     "\n",
     "# set up fields\n",
-    "TEXT = data.Field(lower=True, include_lengths=True, batch_first=True, )\n",
+    "TEXT = data.Field(lower=True, batch_first=True, )\n",
     "LABEL = data.Field(sequential=False, unk_token=None)\n",
     "\n",
     "# make splits for data\n",
@@ -149,10 +149,131 @@
     "    (train, test), batch_sizes=(32, 64), device='cpu')"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch.utils.data"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 7,
    "metadata": {},
+   "outputs": [],
+   "source": [
+    "class TorchDataset(torch.utils.data.Dataset):\n",
+    "    def __init__(self, dataset, TEXT, LABEL):\n",
+    "        self.dataset = dataset\n",
+    "        self.TEXT = TEXT\n",
+    "        self.LABEL = LABEL\n",
+    "    \n",
+    "    def __getitem__(self, idx):\n",
+    "        example = self.dataset.examples[idx]\n",
+    "        X = TEXT.numericalize([example.text]).squeeze() # get rid of \"batch\" dimension\n",
+    "        y = LABEL.numericalize([example.label])\n",
+    "        return X, y\n",
+    "    \n",
+    "    def __len__(self):\n",
+    "        return len(self.dataset)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_dataset = TorchDataset(train, TEXT, LABEL)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "e1 = train_dataset[0]\n",
+    "e2 = train_dataset[1]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from torch.nn.utils.rnn import pad_sequence"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "torch.Size([136])"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "e1[0].shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "torch.Size([143])"
+      ]
+     },
+     "execution_count": 12,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "e2[0].shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "torch.Size([2, 143])"
+      ]
+     },
+     "execution_count": 13,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# hoping skorch will unzip the tuple as follows\n",
+    "# padding_value ought to be 1: TEXT.vocab.stoi[TEXT.pad_token]\n",
+    "pad_sequence(list(zip(*[e1, e2]))[0], batch_first=True, padding_value=1).shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
    "outputs": [
     {
      "data": {
@@ -160,7 +281,7 @@
        "['<unk>', '<pad>', 'the', 'a', 'and']"
       ]
      },
-     "execution_count": 7,
+     "execution_count": 14,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -173,7 +294,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -182,7 +303,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 16,
    "metadata": {},
    "outputs": [
     {
@@ -204,7 +325,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
+   "execution_count": 17,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -215,46 +336,41 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 18,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "tensor([[  542,  5101,    41,  ...,     1,     1,     1],\n",
-      "        [   10,     7,    31,  ...,     1,     1,     1],\n",
-      "        [    0,    49, 15842,  ...,     1,     1,     1],\n",
+      "tensor([[  310,     9,    90,  ...,     1,     1,     1],\n",
+      "        [   94,  1236, 10251,  ...,     1,     1,     1],\n",
+      "        [    9,   297,  2432,  ...,     1,     1,     1],\n",
       "        ...,\n",
-      "        [   25,    86,   311,  ...,     1,     1,     1],\n",
-      "        [    9,   157,   168,  ...,     1,     1,     1],\n",
-      "        [   10,    49,   235,  ...,     1,     1,     1]])\n",
-      "tensor([132, 121,  72, 343, 112,  90, 226, 982, 118, 341, 145,  64, 149,  87,\n",
-      "        244, 270, 122,  94, 141,  76, 154, 151, 429,  97, 104,  76, 120, 125,\n",
-      "        120, 118, 134, 154])\n"
+      "        [   46,    25,    37,  ...,     1,     1,     1],\n",
+      "        [   10,   151,    37,  ...,     1,     1,     1],\n",
+      "        [    9,   586,  7573,  ...,     1,     1,     1]])\n"
      ]
     }
    ],
    "source": [
     "# numericalized tokens\n",
-    "print(batch.text[0])\n",
-    "# sequence lengths\n",
-    "print(batch.text[1])"
+    "print(batch.text)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 19,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "tensor([1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0,\n",
-       "        0, 1, 0, 1, 1, 0, 1, 1])"
+       "tensor([0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1,\n",
+       "        0, 1, 1, 1, 0, 1, 0, 1])"
       ]
      },
-     "execution_count": 12,
+     "execution_count": 19,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -265,7 +381,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 20,
    "metadata": {},
    "outputs": [
     {
@@ -274,7 +390,7 @@
        "defaultdict(None, {'neg': 0, 'pos': 1})"
       ]
      },
-     "execution_count": 13,
+     "execution_count": 20,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -294,7 +410,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 14,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -305,47 +421,47 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "tensor([[0.4698, 0.5302],\n",
-       "        [0.3791, 0.6209],\n",
-       "        [0.2715, 0.7285],\n",
-       "        [0.3220, 0.6780],\n",
-       "        [0.3979, 0.6021],\n",
-       "        [0.4197, 0.5803],\n",
-       "        [0.3409, 0.6591],\n",
-       "        [0.2695, 0.7305],\n",
-       "        [0.4400, 0.5600],\n",
-       "        [0.4209, 0.5791],\n",
-       "        [0.4640, 0.5360],\n",
-       "        [0.4475, 0.5525],\n",
-       "        [0.4261, 0.5739],\n",
-       "        [0.3032, 0.6968],\n",
-       "        [0.3452, 0.6548],\n",
-       "        [0.3803, 0.6197],\n",
-       "        [0.3741, 0.6259],\n",
-       "        [0.3796, 0.6204],\n",
-       "        [0.5005, 0.4995],\n",
-       "        [0.3706, 0.6294],\n",
-       "        [0.3142, 0.6858],\n",
-       "        [0.3772, 0.6228],\n",
-       "        [0.3281, 0.6719],\n",
-       "        [0.3190, 0.6810],\n",
-       "        [0.3340, 0.6660],\n",
-       "        [0.3948, 0.6052],\n",
-       "        [0.2351, 0.7649],\n",
-       "        [0.4538, 0.5462],\n",
-       "        [0.2321, 0.7679],\n",
-       "        [0.4226, 0.5774],\n",
-       "        [0.2449, 0.7551],\n",
-       "        [0.3471, 0.6529]], device='cuda:0', grad_fn=<SoftmaxBackward>)"
+       "tensor([[0.4948, 0.5052],\n",
+       "        [0.3819, 0.6181],\n",
+       "        [0.2845, 0.7155],\n",
+       "        [0.3763, 0.6237],\n",
+       "        [0.4416, 0.5584],\n",
+       "        [0.4124, 0.5876],\n",
+       "        [0.3489, 0.6511],\n",
+       "        [0.3170, 0.6830],\n",
+       "        [0.4920, 0.5080],\n",
+       "        [0.4617, 0.5383],\n",
+       "        [0.4359, 0.5641],\n",
+       "        [0.3998, 0.6002],\n",
+       "        [0.4126, 0.5874],\n",
+       "        [0.3464, 0.6536],\n",
+       "        [0.3516, 0.6484],\n",
+       "        [0.3844, 0.6156],\n",
+       "        [0.3831, 0.6169],\n",
+       "        [0.3435, 0.6565],\n",
+       "        [0.4960, 0.5040],\n",
+       "        [0.3020, 0.6980],\n",
+       "        [0.2781, 0.7219],\n",
+       "        [0.3688, 0.6312],\n",
+       "        [0.3789, 0.6211],\n",
+       "        [0.2984, 0.7016],\n",
+       "        [0.3126, 0.6874],\n",
+       "        [0.4026, 0.5974],\n",
+       "        [0.1881, 0.8119],\n",
+       "        [0.4694, 0.5306],\n",
+       "        [0.2728, 0.7272],\n",
+       "        [0.5161, 0.4839],\n",
+       "        [0.2169, 0.7831],\n",
+       "        [0.4119, 0.5881]], device='cuda:0', grad_fn=<ExpBackward>)"
       ]
      },
-     "execution_count": 15,
+     "execution_count": 22,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -353,13 +469,14 @@
    "source": [
     "# smoketest\n",
     "model = CNN(pretrained_embeddings=TEXT.vocab.vectors).to(device)\n",
-    "gpu_batch = batch.text[0].to(device)\n",
-    "model(gpu_batch)"
+    "gpu_batch = batch.text.to(device)\n",
+    "model_out = model(gpu_batch)\n",
+    "torch.exp(model_out)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -368,7 +485,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -377,7 +494,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -393,18 +510,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 41,
    "metadata": {},
    "outputs": [],
    "source": [
     "import skorch\n",
-    "from skorch import NeuralNetClassifier\n",
+    "from skorch import NeuralNet, NeuralNetClassifier \n",
     "import torch.optim as optim"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 27,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -463,7 +580,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 28,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -472,47 +589,98 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 29,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from torch.utils.data import Dataset, DataLoader"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 30,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def pad_batch(batch):\n",
+    "    text, label = list(zip(*batch))\n",
+    "    padded_batch = pad_sequence(text, batch_first=True, padding_value=1)\n",
+    "    return padded_batch, torch.cat(label)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
    "metadata": {},
    "outputs": [],
    "source": [
+    "from skorch.callbacks import EpochScoring"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "accuracy = EpochScoring('accuracy', lower_is_better=False)\n",
+    "callbacks = [accuracy]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 38,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# if you want to use a custom DataLoader, you must use NeuralNet\n",
+    "# also, not immediately obvious that for NeuralNet you are responsible for applying the log function\n",
+    "# whereas for NeuralNetClassifier, you are not\n",
+    "\n",
     "# NB: not ideal to be using softmax + log + NLLLoss\n",
     "# see discussion: https://github.com/skorch-dev/skorch/issues/637\n",
-    "skorch_model = NeuralNetClassifier(\n",
+    "skorch_model = NeuralNet(\n",
     "                CNN,\n",
     "                device=device,\n",
     "                max_epochs=2,\n",
     "                lr=0.001,\n",
     "                optimizer=optim.Adam,\n",
     "                criterion=nn.NLLLoss,\n",
+    "                iterator_train__collate_fn=pad_batch,\n",
+    "                iterator_train__shuffle=True,\n",
+    "                iterator_valid__collate_fn=pad_batch,\n",
+    "                iterator_valid__shuffle=False,\n",
     "                train_split=skorch.dataset.CVSplit(.2), # NB: this witholds 20% of the training data for validation\n",
     "                module__n_filters=100,\n",
     "                module__filter_sizes=(2,3,4),\n",
     "                module__dropout=0.2,\n",
     "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
     "                batch_size=32,\n",
-    "                verbose=2)"
+    "                verbose=2)\n",
+    "# getting the following error when trying to compute accuracy\n",
+    "# ValueError: Classification metrics can't handle a mix of binary and continuous-multioutput targets\n",
+    "#                 callbacks=callbacks)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": 39,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "  epoch    train_loss    valid_acc    valid_loss      dur\n",
-      "-------  ------------  -----------  ------------  -------\n",
-      "      1        \u001b[36m0.4689\u001b[0m       \u001b[32m0.8370\u001b[0m        \u001b[35m0.3656\u001b[0m  60.5744\n",
-      "      2        \u001b[36m0.3466\u001b[0m       \u001b[32m0.8524\u001b[0m        \u001b[35m0.3413\u001b[0m  58.6526\n"
+      "  epoch    train_loss    valid_loss      dur\n",
+      "-------  ------------  ------------  -------\n",
+      "      1        \u001b[36m0.4723\u001b[0m        \u001b[32m0.3617\u001b[0m  20.1019\n",
+      "      2        \u001b[36m0.3534\u001b[0m        \u001b[32m0.3248\u001b[0m  29.7210\n"
      ]
     },
     {
      "data": {
       "text/plain": [
-       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
+       "<class 'skorch.net.NeuralNet'>[initialized](\n",
        "  module_=CNN(\n",
        "    (embedding): Embedding(25002, 100)\n",
        "    (conv_0): Conv1d(1, 100, kernel_size=(2, 100), stride=(1,))\n",
@@ -524,13 +692,45 @@
        ")"
       ]
      },
-     "execution_count": 25,
+     "execution_count": 39,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# NB: now that we implemented our custom batching we reduced compute time from 60 seconds\n",
+    "# per epoch to 20 seconds per epoch. That's huge.\n",
+    "# ((60 - 20) / 60)*100 = ~66%\n",
+    "skorch_model.fit(train_dataset)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 40,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "0.6666666666666666"
+      ]
+     },
+     "execution_count": 40,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "skorch_model.fit(X_train, y_train)"
+    "(60 - 20) / 60 "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "skorch_model."
    ]
   },
   {

From 8a15332c885dc7be9e38a8e066efd2f676dcc084 Mon Sep 17 00:00:00 2001
From: Todd Morrill <tmorrill12@gmail.com>
Date: Mon, 25 May 2020 10:32:22 -0400
Subject: [PATCH 4/6] skorch running well, hyperband still throwing errors

---
 machine-learning/model.py                    |   14 +-
 machine-learning/skorch-hyperparam-opt.ipynb | 1542 ++++++++----------
 2 files changed, 723 insertions(+), 833 deletions(-)

diff --git a/machine-learning/model.py b/machine-learning/model.py
index 7ad44ea4..6f48270d 100644
--- a/machine-learning/model.py
+++ b/machine-learning/model.py
@@ -4,12 +4,14 @@
 import torch.nn.functional as F
 from torch.nn.utils.rnn import pad_sequence
 import torchtext
+import numpy as np
 
 
 class CNN(nn.Module):
-    def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.2, pretrained_embeddings=None):
+    def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.2, pretrained_embeddings=None, TEXT=None):
         
         super().__init__()
+        self.TEXT = TEXT
         # will be used to initialize model embeddings layer
         self.embedding = nn.Embedding.from_pretrained(pretrained_embeddings)
         self.embedding.weight.requires_grad = False # save some computation
@@ -27,10 +29,12 @@ def __init__(self, n_filters=100, filter_sizes=(2,3,4), output_dim=2, dropout=0.
         self.dropout = nn.Dropout(dropout)
         
     def forward(self, text):
-        # padding_value ought to be 1: TEXT.vocab.stoi[TEXT.pad_token]
-        padded_batch = pad_sequence(text, batch_first=True, padding_value=1)
+#         # bit of a hack to preprocess data inside the network
+#         if isinstance(text, np.ndarray):
+#             text = self.TEXT.process(text)
+        
         #text = [batch size, sent len]
-        embedded = self.embedding(padded_batch)
+        embedded = self.embedding(text)
         #embedded = [batch size, sent len, emb dim]
         embedded = embedded.unsqueeze(1)
         #embedded = [batch size, 1, sent len, emb dim]
@@ -46,4 +50,4 @@ def forward(self, text):
         #cat = [batch size, n_filters * len(filter_sizes)]
         logits = self.fc(cat)
         #logits = [batch_size, output_dim]
-        return torch.log(F.softmax(logits, dim=-1))
\ No newline at end of file
+        return F.softmax(logits, dim=-1)
\ No newline at end of file
diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
index 69f6d639..c06d1443 100644
--- a/machine-learning/skorch-hyperparam-opt.ipynb
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -26,7 +26,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
@@ -37,7 +37,7 @@
        "<td style=\"vertical-align: top; border: 0px solid white\">\n",
        "<h3 style=\"text-align: left;\">Client</h3>\n",
        "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
-       "  <li><b>Scheduler: </b>tcp://127.0.0.1:41949</li>\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42917</li>\n",
        "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
        "</ul>\n",
        "</td>\n",
@@ -53,10 +53,10 @@
        "</table>"
       ],
       "text/plain": [
-       "<Client: 'tcp://127.0.0.1:41949' processes=1 threads=1, memory=33.68 GB>"
+       "<Client: 'tcp://127.0.0.1:42917' processes=1 threads=1, memory=33.68 GB>"
       ]
      },
-     "execution_count": 1,
+     "execution_count": 2,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -78,7 +78,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -103,18 +103,20 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
     "import torchtext\n",
     "from torchtext import data\n",
-    "from torchtext import datasets"
+    "from torchtext import datasets\n",
+    "from torch.utils.data import Dataset, DataLoader\n",
+    "from torch.nn.utils.rnn import pad_sequence"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -123,7 +125,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -144,49 +146,54 @@
     "TEXT.build_vocab(train, vectors=vocab, max_size=max_size)\n",
     "LABEL.build_vocab(train)\n",
     "\n",
+    "# sadly I don't think we can use this \n",
     "# make iterator for splits\n",
-    "train_iter, test_iter = data.BucketIterator.splits(\n",
-    "    (train, test), batch_sizes=(32, 64), device='cpu')"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 6,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import torch.utils.data"
+    "# train_iter, test_iter = data.BucketIterator.splits(\n",
+    "#     (train, test), batch_sizes=(32, 64), device='cpu')"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 7,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['<unk>', '<pad>', 'the', 'a', 'and']"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
-    "class TorchDataset(torch.utils.data.Dataset):\n",
-    "    def __init__(self, dataset, TEXT, LABEL):\n",
-    "        self.dataset = dataset\n",
-    "        self.TEXT = TEXT\n",
-    "        self.LABEL = LABEL\n",
-    "    \n",
-    "    def __getitem__(self, idx):\n",
-    "        example = self.dataset.examples[idx]\n",
-    "        X = TEXT.numericalize([example.text]).squeeze() # get rid of \"batch\" dimension\n",
-    "        y = LABEL.numericalize([example.label])\n",
-    "        return X, y\n",
-    "    \n",
-    "    def __len__(self):\n",
-    "        return len(self.dataset)"
+    "# itos := index-to-string\n",
+    "# note the 2 extra tokens added for us: '<unk>', '<pad>'\n",
+    "TEXT.vocab.itos[:5]"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 8,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "defaultdict(None, {'neg': 0, 'pos': 1})"
+      ]
+     },
+     "execution_count": 8,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
-    "train_dataset = TorchDataset(train, TEXT, LABEL)"
+    "# stoi := string-to-index\n",
+    "# check on the meaning of these zeroes and ones\n",
+    "LABEL.vocab.stoi"
    ]
   },
   {
@@ -195,37 +202,47 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "e1 = train_dataset[0]\n",
-    "e2 = train_dataset[1]"
+    "assert (len(TEXT.vocab.itos) == max_size + 2)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 10,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.']\n",
+      "\n",
+      "pos\n"
+     ]
+    }
+   ],
    "source": [
-    "from torch.nn.utils.rnn import pad_sequence"
+    "# peek at the data\n",
+    "print(train.examples[0].text)\n",
+    "print()\n",
+    "print(train.examples[0].label)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 11,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "torch.Size([136])"
-      ]
-     },
-     "execution_count": 11,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "e1[0].shape"
+    "# if you could use data.BucketIterator.splits\n",
+    "\n",
+    "# sadly train_iter isn't actually an iter..\n",
+    "# peek at a batch of data\n",
+    "# batch = next(iter(train_iter))\n",
+    "\n",
+    "# numericalized tokens\n",
+    "# print(batch.text)\n",
+    "\n",
+    "# print(batch.label)"
    ]
   },
   {
@@ -236,7 +253,7 @@
     {
      "data": {
       "text/plain": [
-       "torch.Size([143])"
+       "torch.Size([2, 143])"
       ]
      },
      "execution_count": 12,
@@ -245,51 +262,49 @@
     }
    ],
    "source": [
-    "e2[0].shape"
+    "TEXT.process([train[0].text, train[1].text]).shape"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 13,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "torch.Size([2, 143])"
-      ]
-     },
-     "execution_count": 13,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "# hoping skorch will unzip the tuple as follows\n",
-    "# padding_value ought to be 1: TEXT.vocab.stoi[TEXT.pad_token]\n",
-    "pad_sequence(list(zip(*[e1, e2]))[0], batch_first=True, padding_value=1).shape"
+    "# custom dataset class required to work with Skorch\n",
+    "class TorchDataset(Dataset):\n",
+    "    def __init__(self, dataset):\n",
+    "        self.dataset = dataset\n",
+    "\n",
+    "    def __getitem__(self, idx):\n",
+    "        example = self.dataset.examples[idx]\n",
+    "        return example.text, example.label\n",
+    "    \n",
+    "    def __len__(self):\n",
+    "        return len(self.dataset)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 14,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "['<unk>', '<pad>', 'the', 'a', 'and']"
-      ]
-     },
-     "execution_count": 14,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "# itos := index-to-string\n",
-    "# note the 2 extra tokens added for us: '<unk>', '<pad>'\n",
-    "TEXT.vocab.itos[:5]"
+    "# # custom dataset class required to work with Skorch\n",
+    "# class TorchDataset(torch.utils.data.Dataset):\n",
+    "#     def __init__(self, dataset, TEXT, LABEL):\n",
+    "#         self.dataset = dataset\n",
+    "#         self.TEXT = TEXT\n",
+    "#         self.LABEL = LABEL\n",
+    "    \n",
+    "#     def __getitem__(self, idx):\n",
+    "#         example = self.dataset.examples[idx]\n",
+    "#         X = TEXT.numericalize([example.text]).squeeze() # get rid of \"batch\" dimension\n",
+    "#         y = LABEL.numericalize([example.label])\n",
+    "#         return X, y\n",
+    "    \n",
+    "#     def __len__(self):\n",
+    "#         return len(self.dataset)"
    ]
   },
   {
@@ -298,7 +313,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "assert (len(TEXT.vocab.itos) == max_size + 2)"
+    "ex = train.examples[0]"
    ]
   },
   {
@@ -307,20 +322,18 @@
    "metadata": {},
    "outputs": [
     {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.']\n",
-      "\n",
-      "pos\n"
-     ]
+     "data": {
+      "text/plain": [
+       "'a masterful treatment of james caine\\'s \"the postman always rings twice\" as luchino visconti\\'s first film shot primarily around ferrara in a soulless war-torn italy. the original negative was thought destroyed but visconti saved a print and fortunately we can see this early neo-realist work today. a ruggedly handsome massimo girotti and clara calamai (who had recently revealed her breasts in la cena delle beffe\" (1941), star as the sensually-charged and ill-fated lovers who plot to kill her husband. unusual ending in which, although crime does not pay, one pays in a way not directly linked to the crime. excellent direction, script, acting, and cinematography. reportedly not as good as the french \"le dernier tournant\\' (1939) but probably better than the us version (1946) featuring lana turner and john garfield in the lead roles. highly recommended.'"
+      ]
+     },
+     "execution_count": 16,
+     "metadata": {},
+     "output_type": "execute_result"
     }
    ],
    "source": [
-    "# peek at the data\n",
-    "print(train.examples[0].text)\n",
-    "print()\n",
-    "print(train.examples[0].label)"
+    "' '.join(ex.text)"
    ]
   },
   {
@@ -329,9 +342,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# sadly train_iter isn't actually an iter..\n",
-    "# peek at a batch of data\n",
-    "batch = next(iter(train_iter))"
+    "train_dataset = TorchDataset(train)\n",
+    "test_dataset = TorchDataset(test)"
    ]
   },
   {
@@ -343,62 +355,48 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "tensor([[  310,     9,    90,  ...,     1,     1,     1],\n",
-      "        [   94,  1236, 10251,  ...,     1,     1,     1],\n",
-      "        [    9,   297,  2432,  ...,     1,     1,     1],\n",
-      "        ...,\n",
-      "        [   46,    25,    37,  ...,     1,     1,     1],\n",
-      "        [   10,   151,    37,  ...,     1,     1,     1],\n",
-      "        [    9,   586,  7573,  ...,     1,     1,     1]])\n"
+      "(['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.'], 'pos')\n"
      ]
     }
    ],
    "source": [
-    "# numericalized tokens\n",
-    "print(batch.text)"
+    "print(train_dataset[0])"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 19,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "tensor([0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1,\n",
-       "        0, 1, 1, 1, 0, 1, 0, 1])"
-      ]
-     },
-     "execution_count": 19,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "batch.label"
+    "def pad_batch(batch, TEXT, LABEL):\n",
+    "    text, label = list(zip(*batch))\n",
+    "    # numericalized and padded text representation\n",
+    "    text_processed = TEXT.process(text)\n",
+    "    label_processed = LABEL.process(label)\n",
+    "    return text_processed, label_processed\n",
+    "\n",
+    "from functools import partial\n",
+    "\n",
+    "pad_batch_partial = partial(pad_batch, TEXT=TEXT, LABEL=LABEL)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 20,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "defaultdict(None, {'neg': 0, 'pos': 1})"
-      ]
-     },
-     "execution_count": 20,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "# stoi := string-to-index\n",
-    "# check on the meaning of these zeroes and ones\n",
-    "LABEL.vocab.stoi"
+    "train_dataloader = DataLoader(train_dataset, batch_size=32, shuffle=True, collate_fn=pad_batch_partial)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "batch = next(iter(train_dataloader))"
    ]
   },
   {
@@ -410,7 +408,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -421,62 +419,59 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [
     {
-     "data": {
-      "text/plain": [
-       "tensor([[0.4948, 0.5052],\n",
-       "        [0.3819, 0.6181],\n",
-       "        [0.2845, 0.7155],\n",
-       "        [0.3763, 0.6237],\n",
-       "        [0.4416, 0.5584],\n",
-       "        [0.4124, 0.5876],\n",
-       "        [0.3489, 0.6511],\n",
-       "        [0.3170, 0.6830],\n",
-       "        [0.4920, 0.5080],\n",
-       "        [0.4617, 0.5383],\n",
-       "        [0.4359, 0.5641],\n",
-       "        [0.3998, 0.6002],\n",
-       "        [0.4126, 0.5874],\n",
-       "        [0.3464, 0.6536],\n",
-       "        [0.3516, 0.6484],\n",
-       "        [0.3844, 0.6156],\n",
-       "        [0.3831, 0.6169],\n",
-       "        [0.3435, 0.6565],\n",
-       "        [0.4960, 0.5040],\n",
-       "        [0.3020, 0.6980],\n",
-       "        [0.2781, 0.7219],\n",
-       "        [0.3688, 0.6312],\n",
-       "        [0.3789, 0.6211],\n",
-       "        [0.2984, 0.7016],\n",
-       "        [0.3126, 0.6874],\n",
-       "        [0.4026, 0.5974],\n",
-       "        [0.1881, 0.8119],\n",
-       "        [0.4694, 0.5306],\n",
-       "        [0.2728, 0.7272],\n",
-       "        [0.5161, 0.4839],\n",
-       "        [0.2169, 0.7831],\n",
-       "        [0.4119, 0.5881]], device='cuda:0', grad_fn=<ExpBackward>)"
-      ]
-     },
-     "execution_count": 22,
-     "metadata": {},
-     "output_type": "execute_result"
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor([[0.5732, 0.4268],\n",
+      "        [0.5372, 0.4628],\n",
+      "        [0.5615, 0.4385],\n",
+      "        [0.6806, 0.3194],\n",
+      "        [0.6352, 0.3648],\n",
+      "        [0.5678, 0.4322],\n",
+      "        [0.6505, 0.3495],\n",
+      "        [0.4325, 0.5675],\n",
+      "        [0.6019, 0.3981],\n",
+      "        [0.5857, 0.4143],\n",
+      "        [0.6010, 0.3990],\n",
+      "        [0.5662, 0.4338],\n",
+      "        [0.6494, 0.3506],\n",
+      "        [0.4390, 0.5610],\n",
+      "        [0.6041, 0.3959],\n",
+      "        [0.7121, 0.2879],\n",
+      "        [0.5781, 0.4219],\n",
+      "        [0.5727, 0.4273],\n",
+      "        [0.5169, 0.4831],\n",
+      "        [0.6531, 0.3469],\n",
+      "        [0.5080, 0.4920],\n",
+      "        [0.5611, 0.4389],\n",
+      "        [0.5560, 0.4440],\n",
+      "        [0.6448, 0.3552],\n",
+      "        [0.5697, 0.4303],\n",
+      "        [0.5798, 0.4202],\n",
+      "        [0.6656, 0.3344],\n",
+      "        [0.4933, 0.5067],\n",
+      "        [0.5143, 0.4857],\n",
+      "        [0.5518, 0.4482],\n",
+      "        [0.5806, 0.4194],\n",
+      "        [0.6070, 0.3930]], device='cuda:0', grad_fn=<SoftmaxBackward>)\n"
+     ]
     }
    ],
    "source": [
     "# smoketest\n",
     "model = CNN(pretrained_embeddings=TEXT.vocab.vectors).to(device)\n",
-    "gpu_batch = batch.text.to(device)\n",
+    "gpu_batch = batch[0].to(device)\n",
     "model_out = model(gpu_batch)\n",
-    "torch.exp(model_out)"
+    "print(model_out)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -485,7 +480,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -494,7 +489,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": 26,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -510,186 +505,216 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 27,
    "metadata": {},
    "outputs": [],
    "source": [
     "import skorch\n",
-    "from skorch import NeuralNet, NeuralNetClassifier \n",
+    "from skorch import NeuralNetClassifier \n",
     "import torch.optim as optim"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": 28,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# this is a really unfortunate hack to make torchtext batching semantics work with skorch and dask\n",
-    "# the downside here is that we're no longer padding to the longest sequence in the batch, rather\n",
-    "# we're padding to the longest sequence in the *dataset*, which results in signifcantly more\n",
-    "# computation and thus significantly more time to train a model\n",
-    "# of course, you could set a max sequence length but that's not an ideal solution\n",
-    "# another solution would be to create a different dataset object, but then you can't use torchtext,\n",
-    "# which really is quite handy\n",
-    "\n",
-    "# train=True shuffles the data\n",
-    "train_iter_skorch = torchtext.data.Iterator(train, batch_size=len(train), train=True, sort=False, device='cpu')\n",
-    "test_iter_skorch = torchtext.data.Iterator(test, batch_size=len(test), train=False, sort=False, device='cpu')"
+    "import torch.nn as nn"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
+   "execution_count": 29,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# takes some time to numericalize the whole dataset\n",
-    "\n",
-    "# also notice that skorch and dask expect numpy arrays, which isn't ideal since it ties you to the cpu.\n",
-    "# meanwhile, projects like https://rapids.ai/ are moving toward all GPU computation, avoiding the cpu altogether.\n",
-    "for batch in train_iter_skorch:\n",
-    "    X_train = batch.text[0].numpy()\n",
-    "    y_train = batch.label.numpy()\n",
+    "# if you want to use a custom DataLoader, you must use NeuralNet\n",
+    "# also, not immediately obvious that for NeuralNet you are responsible for applying the log function\n",
+    "# whereas for NeuralNetClassifier, you are not\n",
     "\n",
-    "for batch in test_iter_skorch:\n",
-    "    X_test = batch.text[0].numpy()\n",
-    "    y_test = batch.label.numpy()"
+    "# NB: not ideal to be using softmax + log + NLLLoss\n",
+    "# see discussion: https://github.com/skorch-dev/skorch/issues/637\n",
+    "skorch_model = NeuralNetClassifier(\n",
+    "                CNN,\n",
+    "                device=device,\n",
+    "                max_epochs=2,\n",
+    "                lr=0.001,\n",
+    "                optimizer=optim.Adam,\n",
+    "                criterion=nn.NLLLoss,\n",
+    "                iterator_train=DataLoader,\n",
+    "                iterator_train__shuffle=True,\n",
+    "                iterator_train__batch_size=32,\n",
+    "                iterator_train__collate_fn=pad_batch_partial,\n",
+    "                iterator_train__num_workers=8,\n",
+    "                iterator_valid=DataLoader,\n",
+    "                iterator_valid__shuffle=False,\n",
+    "                iterator_valid__batch_size=64,\n",
+    "                iterator_valid__collate_fn=pad_batch_partial,\n",
+    "                iterator_valid__num_workers=8,\n",
+    "                train_split=skorch.dataset.CVSplit(.2), # NB: this witholds 20% of the training data for validation\n",
+    "                module__n_filters=100,\n",
+    "                module__filter_sizes=(2,3,4),\n",
+    "                module__dropout=0.2,\n",
+    "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
+    "                verbose=2)\n",
+    "# getting the following error when trying to compute accuracy\n",
+    "# ValueError: Classification metrics can't handle a mix of binary and continuous-multioutput targets\n",
+    "#                 callbacks=callbacks)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "  epoch    train_loss    valid_acc    valid_loss      dur\n",
+      "-------  ------------  -----------  ------------  -------\n",
+      "      1        \u001b[36m0.4630\u001b[0m       \u001b[32m0.8312\u001b[0m        \u001b[35m0.3808\u001b[0m  33.8281\n",
+      "      2        \u001b[36m0.3454\u001b[0m       \u001b[32m0.8540\u001b[0m        \u001b[35m0.3361\u001b[0m  34.1240\n"
+     ]
+    },
     {
      "data": {
       "text/plain": [
-       "(25000, 2470)"
+       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
+       "  module_=CNN(\n",
+       "    (embedding): Embedding(25002, 100)\n",
+       "    (conv_0): Conv1d(1, 100, kernel_size=(2, 100), stride=(1,))\n",
+       "    (conv_1): Conv1d(1, 100, kernel_size=(3, 100), stride=(1,))\n",
+       "    (conv_2): Conv1d(1, 100, kernel_size=(4, 100), stride=(1,))\n",
+       "    (fc): Linear(in_features=300, out_features=2, bias=True)\n",
+       "    (dropout): Dropout(p=0.2, inplace=False)\n",
+       "  ),\n",
+       ")"
       ]
      },
-     "execution_count": 22,
+     "execution_count": 30,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "# notice how awfully large the second dimension is\n",
-    "X_train.shape"
+    "# NB: now that we implemented our custom batching we reduced compute time from 60 seconds\n",
+    "# per epoch to 20 seconds per epoch. That's huge.\n",
+    "# ((60 - 20) / 60)*100 = ~66%\n",
+    "skorch_model.fit(train_dataset, y=None)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 28,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [],
    "source": [
-    "import torch.nn as nn"
+    "# this isn't working as expected\n",
+    "# skorch_model.score(test_dataset)\n",
+    "# TypeError: score() missing 1 required positional argument: 'y'\n",
+    "# skorch_model.score(test_dataset, y=None)\n",
+    "# ValueError: Expected array-like (array or non-string sequence), got None"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 32,
    "metadata": {},
    "outputs": [],
    "source": [
-    "from torch.utils.data import Dataset, DataLoader"
+    "# score manually\n",
+    "test_dataloader = DataLoader(test_dataset, batch_size=len(test_dataset), shuffle=False, collate_fn=pad_batch_partial, num_workers=8)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 33,
    "metadata": {},
    "outputs": [],
    "source": [
-    "def pad_batch(batch):\n",
-    "    text, label = list(zip(*batch))\n",
-    "    padded_batch = pad_sequence(text, batch_first=True, padding_value=1)\n",
-    "    return padded_batch, torch.cat(label)"
+    "# sadly processing the entire test set twice just to score the model\n",
+    "processed_test_data = next(iter(test_dataloader))"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 34,
    "metadata": {},
    "outputs": [],
    "source": [
-    "from skorch.callbacks import EpochScoring"
+    "# quick check on the test set accuracy\n",
+    "test_preds = skorch_model.predict(test_dataset)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 35,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([0, 1, 1, ..., 0, 0, 0])"
+      ]
+     },
+     "execution_count": 35,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "test_preds"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 36,
    "metadata": {},
    "outputs": [],
    "source": [
-    "accuracy = EpochScoring('accuracy', lower_is_better=False)\n",
-    "callbacks = [accuracy]"
+    "test_labels = processed_test_data[1].numpy()"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 37,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# if you want to use a custom DataLoader, you must use NeuralNet\n",
-    "# also, not immediately obvious that for NeuralNet you are responsible for applying the log function\n",
-    "# whereas for NeuralNetClassifier, you are not\n",
-    "\n",
-    "# NB: not ideal to be using softmax + log + NLLLoss\n",
-    "# see discussion: https://github.com/skorch-dev/skorch/issues/637\n",
-    "skorch_model = NeuralNet(\n",
-    "                CNN,\n",
-    "                device=device,\n",
-    "                max_epochs=2,\n",
-    "                lr=0.001,\n",
-    "                optimizer=optim.Adam,\n",
-    "                criterion=nn.NLLLoss,\n",
-    "                iterator_train__collate_fn=pad_batch,\n",
-    "                iterator_train__shuffle=True,\n",
-    "                iterator_valid__collate_fn=pad_batch,\n",
-    "                iterator_valid__shuffle=False,\n",
-    "                train_split=skorch.dataset.CVSplit(.2), # NB: this witholds 20% of the training data for validation\n",
-    "                module__n_filters=100,\n",
-    "                module__filter_sizes=(2,3,4),\n",
-    "                module__dropout=0.2,\n",
-    "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
-    "                batch_size=32,\n",
-    "                verbose=2)\n",
-    "# getting the following error when trying to compute accuracy\n",
-    "# ValueError: Classification metrics can't handle a mix of binary and continuous-multioutput targets\n",
-    "#                 callbacks=callbacks)"
+    "from sklearn.metrics import accuracy_score"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 39,
+   "execution_count": 38,
    "metadata": {},
    "outputs": [
     {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "  epoch    train_loss    valid_loss      dur\n",
-      "-------  ------------  ------------  -------\n",
-      "      1        \u001b[36m0.4723\u001b[0m        \u001b[32m0.3617\u001b[0m  20.1019\n",
-      "      2        \u001b[36m0.3534\u001b[0m        \u001b[32m0.3248\u001b[0m  29.7210\n"
-     ]
-    },
+     "data": {
+      "text/plain": [
+       "0.85048"
+      ]
+     },
+     "execution_count": 38,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "accuracy_score(test_labels, test_preds)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 39,
+   "metadata": {},
+   "outputs": [
     {
      "data": {
       "text/plain": [
-       "<class 'skorch.net.NeuralNet'>[initialized](\n",
-       "  module_=CNN(\n",
-       "    (embedding): Embedding(25002, 100)\n",
-       "    (conv_0): Conv1d(1, 100, kernel_size=(2, 100), stride=(1,))\n",
-       "    (conv_1): Conv1d(1, 100, kernel_size=(3, 100), stride=(1,))\n",
-       "    (conv_2): Conv1d(1, 100, kernel_size=(4, 100), stride=(1,))\n",
-       "    (fc): Linear(in_features=300, out_features=2, bias=True)\n",
-       "    (dropout): Dropout(p=0.2, inplace=False)\n",
-       "  ),\n",
-       ")"
+       "(array([0, 1]), array([12500, 12500]))"
       ]
      },
      "execution_count": 39,
@@ -698,105 +723,236 @@
     }
    ],
    "source": [
-    "# NB: now that we implemented our custom batching we reduced compute time from 60 seconds\n",
-    "# per epoch to 20 seconds per epoch. That's huge.\n",
-    "# ((60 - 20) / 60)*100 = ~66%\n",
-    "skorch_model.fit(train_dataset)"
+    "# random guessing would 50% accuracy so the model is indeed training\n",
+    "np.unique(test_labels, return_counts=True)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 40,
    "metadata": {},
+   "outputs": [],
+   "source": [
+    "# NB: this has no effect on GPU memory usage. If I keyboard interrupt, the workers get\n",
+    "# restarted and memory usage goes down. Deleting these \"handler\" objects doesn't delete\n",
+    "# GPU memory references on the workers. \n",
+    "del skorch_model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Grid search with Hyperband"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 41,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# try to store the raw text in dask arrays and handle preprocessing inside the network\n",
+    "# this isn't ideal because feature engineering really should be separate from modeling"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# raw text\n",
+    "train_text = []\n",
+    "train_label = []\n",
+    "for e in train.examples:\n",
+    "    tokenized_text = e.text\n",
+    "    text_string = ' '.join(tokenized_text)\n",
+    "    train_text.append(text_string)\n",
+    "    label = e.label\n",
+    "    # label = LABEL.process([e.label]).numpy()\n",
+    "    train_label.append(label)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "0.6666666666666666"
+       "['a masterful treatment of james caine\\'s \"the postman always rings twice\" as luchino visconti\\'s first film shot primarily around ferrara in a soulless war-torn italy. the original negative was thought destroyed but visconti saved a print and fortunately we can see this early neo-realist work today. a ruggedly handsome massimo girotti and clara calamai (who had recently revealed her breasts in la cena delle beffe\" (1941), star as the sensually-charged and ill-fated lovers who plot to kill her husband. unusual ending in which, although crime does not pay, one pays in a way not directly linked to the crime. excellent direction, script, acting, and cinematography. reportedly not as good as the french \"le dernier tournant\\' (1939) but probably better than the us version (1946) featuring lana turner and john garfield in the lead roles. highly recommended.']"
       ]
      },
-     "execution_count": 40,
+     "execution_count": 43,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "(60 - 20) / 60 "
+    "train_text[:1]"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 44,
    "metadata": {},
    "outputs": [],
    "source": [
-    "skorch_model."
+    "train_text = np.array(train_text)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 45,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_text = np.expand_dims(train_text, axis=-1)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 46,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "sample_batch = [train_text[0], train_text[10]]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 47,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "sample_text_batch = [i[0] for i in sample_batch]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 48,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "0.85512"
+       "tensor([[    3,     0, 12500,  ...,     1,     1,     1],\n",
+       "        [ 1669,  3651,     9,  ...,  9300,  6517,   411]])"
       ]
      },
-     "execution_count": 26,
+     "execution_count": 48,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "# quick check on the test set accuracy\n",
-    "skorch_model.score(X_test, y_test)"
+    "TEXT.process(sample_text_batch)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": 49,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "(array([0, 1]), array([12500, 12500]))"
+       "torch.Size([2, 858])"
       ]
      },
-     "execution_count": 27,
+     "execution_count": 49,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "# random guessing would 50% accuracy so the model is indeed training\n",
-    "np.unique(y_test, return_counts=True)"
+    "TEXT.process([train_text[0][0], train_text[1][0]]).shape"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 28,
+   "execution_count": 50,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "True"
+      ]
+     },
+     "execution_count": 50,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "isinstance(train_text[:10], np.ndarray)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 51,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# NB: this has no effect on GPU memory usage. If I keyboard interrupt, the workers get\n",
-    "# restarted and memory usage goes down. Deleting these \"handler\" objects doesn't delete\n",
-    "# GPU memory references on the workers. \n",
-    "del skorch_model"
+    "# TEXT.process(train_text[0])"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": 52,
    "metadata": {},
+   "outputs": [],
    "source": [
-    "## Grid search with Hyperband"
+    "train_label = np.array(train_label)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 53,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# # this is a really unfortunate hack to make torchtext batching semantics work with skorch and dask\n",
+    "# # the downside here is that we're no longer padding to the longest sequence in the batch, rather\n",
+    "# # we're padding to the longest sequence in the *dataset*, which results in signifcantly more\n",
+    "# # computation and thus significantly more time to train a model\n",
+    "# # of course, you could set a max sequence length but that's not an ideal solution\n",
+    "# # another solution would be to create a different dataset object, but then you can't use torchtext,\n",
+    "# # which really is quite handy\n",
+    "\n",
+    "# # train=True shuffles the data\n",
+    "# train_iter_skorch = torchtext.data.Iterator(train, batch_size=len(train), train=True, sort=False, device='cpu')\n",
+    "# test_iter_skorch = torchtext.data.Iterator(test, batch_size=len(test), train=False, sort=False, device='cpu')\n",
+    "\n",
+    "# # takes some time to numericalize the whole dataset\n",
+    "\n",
+    "# # also notice that skorch and dask expect numpy arrays, which isn't ideal since it ties you to the cpu.\n",
+    "# # meanwhile, projects like https://rapids.ai/ are moving toward all GPU computation, avoiding the cpu altogether.\n",
+    "# for batch in train_iter_skorch:\n",
+    "#     X_train = batch.text[0].numpy()\n",
+    "#     y_train = batch.label.numpy()\n",
+    "\n",
+    "# for batch in test_iter_skorch:\n",
+    "#     X_test = batch.text[0].numpy()\n",
+    "#     y_test = batch.label.numpy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 54,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# notice how awfully large the second dimension is\n",
+    "# X_train.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 55,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -813,7 +969,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 56,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -826,7 +982,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 57,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -835,7 +991,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 58,
    "metadata": {},
    "outputs": [
     {
@@ -851,30 +1007,30 @@
    "source": [
     "# choose chunk size so that the remainder is not a tiny number\n",
     "print(f'Chunk size: {chunk_size}')\n",
-    "print(f'Total chunks: {math.ceil(len(X_train) / chunk_size)}')\n",
-    "last_chunk_size = len(X_train) % chunk_size\n",
+    "print(f'Total chunks: {math.ceil(len(train) / chunk_size)}')\n",
+    "last_chunk_size = len(train) % chunk_size\n",
     "if last_chunk_size == 0: # i.e. chunk_size evenly divides X_train\n",
     "    last_chunk_size = chunk_size\n",
     "print(f'Last chunk size: {last_chunk_size}')\n",
     "\n",
-    "assert (len(X_train) % chunk_size > 10 or len(X_train) % chunk_size == 0), 'Choose another chunk size'"
+    "assert (len(train) % chunk_size > 10 or len(train) % chunk_size == 0), 'Choose another chunk size'"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 33,
+   "execution_count": 59,
    "metadata": {},
    "outputs": [],
    "source": [
     "import dask.array as da\n",
     "\n",
-    "X = da.from_array(X_train, chunks=(chunk_size, X_train.shape[-1]))\n",
-    "y = da.from_array(y_train, chunks=(chunk_size))"
+    "X = da.from_array(train_text, chunks=(chunk_size))\n",
+    "y = da.from_array(train_label, chunks=(chunk_size))"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 34,
+   "execution_count": 60,
    "metadata": {},
    "outputs": [
     {
@@ -888,41 +1044,41 @@
        "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
-       "    <tr><th> Bytes </th><td> 494.00 MB </td> <td> 247.00 MB </td></tr>\n",
-       "    <tr><th> Shape </th><td> (25000, 2470) </td> <td> (12500, 2470) </td></tr>\n",
+       "    <tr><th> Bytes </th><td> 1.37 GB </td> <td> 685.20 MB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (25000, 1) </td> <td> (12500, 1) </td></tr>\n",
        "    <tr><th> Count </th><td> 3 Tasks </td><td> 2 Chunks </td></tr>\n",
-       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
+       "    <tr><th> Type </th><td> <U13704 </td><td> numpy.ndarray </td></tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</td>\n",
        "<td>\n",
-       "<svg width=\"88\" height=\"170\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "<svg width=\"75\" height=\"170\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
        "\n",
        "  <!-- Horizontal lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"38\" y2=\"0\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"0\" y1=\"60\" x2=\"38\" y2=\"60\" />\n",
-       "  <line x1=\"0\" y1=\"120\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"25\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"60\" x2=\"25\" y2=\"60\" />\n",
+       "  <line x1=\"0\" y1=\"120\" x2=\"25\" y2=\"120\" style=\"stroke-width:2\" />\n",
        "\n",
        "  <!-- Vertical lines -->\n",
        "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"120\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"38\" y1=\"0\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"25\" y1=\"0\" x2=\"25\" y2=\"120\" style=\"stroke-width:2\" />\n",
        "\n",
        "  <!-- Colored Rectangle -->\n",
-       "  <polygon points=\"0.000000,0.000000 38.529848,0.000000 38.529848,120.000000 0.000000,120.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "  <polygon points=\"0.000000,0.000000 25.412617,0.000000 25.412617,120.000000 0.000000,120.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
        "\n",
        "  <!-- Text -->\n",
-       "  <text x=\"19.264924\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >2470</text>\n",
-       "  <text x=\"58.529848\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,58.529848,60.000000)\">25000</text>\n",
+       "  <text x=\"12.706308\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >1</text>\n",
+       "  <text x=\"45.412617\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,45.412617,60.000000)\">25000</text>\n",
        "</svg>\n",
        "</td>\n",
        "</tr>\n",
        "</table>"
       ],
       "text/plain": [
-       "dask.array<array, shape=(25000, 2470), dtype=int64, chunksize=(12500, 2470), chunktype=numpy.ndarray>"
+       "dask.array<array, shape=(25000, 1), dtype=<U13704, chunksize=(12500, 1), chunktype=numpy.ndarray>"
       ]
      },
-     "execution_count": 34,
+     "execution_count": 60,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -933,7 +1089,37 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 35,
+   "execution_count": 61,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# need new collate_fn due to the following error\n",
+    "# ValueError: Expected 2D array, got 1D array instead:\n",
+    "# array=['1' '1' '1' ... '1' '1' '1'].\n",
+    "# Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 62,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def pad_batch_hyperband(batch, TEXT, LABEL):\n",
+    "    text, label = list(zip(*batch))\n",
+    "    # unnecessary extra dimension\n",
+    "    text = [i[0] for i in text]\n",
+    "    # numericalized and padded text representation\n",
+    "    text_processed = TEXT.process(text)\n",
+    "    label_processed = LABEL.process(label)\n",
+    "    return text_processed, label_processed\n",
+    "\n",
+    "pad_batch_hyperband_partial = partial(pad_batch_hyperband, TEXT=TEXT, LABEL=LABEL)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 63,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -944,18 +1130,27 @@
     "                lr=0.001,\n",
     "                optimizer=optim.Adam,\n",
     "                criterion=nn.NLLLoss,\n",
+    "                iterator_train=DataLoader,\n",
+    "                iterator_train__shuffle=True,\n",
+    "                iterator_train__batch_size=32,\n",
+    "                iterator_train__collate_fn=pad_batch_hyperband_partial,\n",
+    "                iterator_valid=DataLoader,\n",
+    "                iterator_valid__collate_fn=pad_batch_hyperband_partial,\n",
+    "                iterator_valid__shuffle=False,\n",
+    "                iterator_valid__batch_size=64,\n",
     "                train_split=None, # let hyperband handle it\n",
     "                module__n_filters=100,\n",
     "                module__filter_sizes=(2, 3, 4),\n",
     "                module__dropout=0.2,\n",
     "                module__pretrained_embeddings=TEXT.vocab.vectors,\n",
+    "                # module__TEXT=TEXT,\n",
     "                batch_size=32,\n",
     "                verbose=2)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 36,
+   "execution_count": 64,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -964,7 +1159,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 37,
+   "execution_count": 65,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -978,7 +1173,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 66,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -987,7 +1182,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 39,
+   "execution_count": 67,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1002,7 +1197,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 40,
+   "execution_count": 68,
    "metadata": {},
    "outputs": [
     {
@@ -1011,7 +1206,7 @@
        "26"
       ]
      },
-     "execution_count": 40,
+     "execution_count": 68,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1022,7 +1217,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 69,
    "metadata": {},
    "outputs": [
     {
@@ -1031,7 +1226,7 @@
        "5"
       ]
      },
-     "execution_count": 41,
+     "execution_count": 69,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1042,7 +1237,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 42,
+   "execution_count": 70,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1051,7 +1246,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 43,
+   "execution_count": 71,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1066,7 +1261,39 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 44,
+   "execution_count": 72,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# I feel like the GPU is spending a lot of time waiting for the CPU to preprocess data\n",
+    "# I've been watching GPU power consumption during training and it's not as high as during\n",
+    "# skorch training, so I tried to increase the number of workers the dataloader uses and\n",
+    "# got the following error: AssertionError: daemonic processes are not allowed to have children"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 83,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'pos'"
+      ]
+     },
+     "execution_count": 83,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "y[0].compute()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 73,
    "metadata": {},
    "outputs": [
     {
@@ -1080,7 +1307,19 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 10.00 MB detected in task graph: \n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/torch/storage.py:34: FutureWarning: pickle support for Storage will be removed in 1.5. Use `torch.save` instead\n",
+      "  warnings.warn(\"pickle support for Storage will be removed in 1.5. Use `torch.save` instead\", FutureWarning)\n",
+      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 23.79 MB detected in task graph: \n",
       "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .0721]]),\n",
       "), 0]\n",
       "Consider scattering large objects ahead of time\n",
@@ -1098,21 +1337,123 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[CV, bracket=0] creating 2 models\n",
+      "[CV, bracket=0] creating 2 models\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
+      "  sk_validation.check_array(sample, *args, **kwargs)\n",
+      "/opt/conda/lib/python3.7/site-packages/torch/storage.py:34: FutureWarning: pickle support for Storage will be removed in 1.5. Use `torch.save` instead\n",
+      "  warnings.warn(\"pickle support for Storage will be removed in 1.5. Use `torch.save` instead\", FutureWarning)\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
       "[CV, bracket=0] For training there are between 10000 and 10000 examples in each chunk\n",
-      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n",
-      "[CV, bracket=0] validation score of 0.8386 received after 1 partial_fit calls\n",
-      "[CV, bracket=1] validation score of 0.8306 received after 1 partial_fit calls\n",
-      "[CV, bracket=0] validation score of 0.8446 received after 8 partial_fit calls\n",
-      "[CV, bracket=1] validation score of 0.8430 received after 2 partial_fit calls\n",
-      "[CV, bracket=1] validation score of 0.8456 received after 6 partial_fit calls\n",
-      "Time to complete grid search: 3813.07 seconds\n"
+      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "tornado.application - ERROR - Multiple exceptions in yield list\n",
+      "Traceback (most recent call last):\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 849, in callback\n",
+      "    result_list.append(f.result())\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
+      "    yielded = self.gen.throw(*exc_info)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 625, in _fit\n",
+      "    prefix=self.prefix,\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
+      "    value = future.result()\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
+      "    yielded = self.gen.throw(*exc_info)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 233, in _fit\n",
+      "    metas = yield client.gather(new_scores)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
+      "    value = future.result()\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/client.py\", line 1826, in _gather\n",
+      "    raise exception.with_traceback(traceback)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 103, in _score\n",
+      "    score = scorer(model, X, y)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/sklearn/metrics/_scorer.py\", line 371, in _passthrough_scorer\n",
+      "    return estimator.score(*args, **kwargs)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/sklearn/base.py\", line 369, in score\n",
+      "    return accuracy_score(y, self.predict(X), sample_weight=sample_weight)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/skorch/classifier.py\", line 210, in predict\n",
+      "    for yp in self.forward_iter(X, training=False):\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/skorch/net.py\", line 917, in forward_iter\n",
+      "    for data in iterator:\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\", line 345, in __next__\n",
+      "    data = self._next_data()\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\", line 385, in _next_data\n",
+      "    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\", line 47, in fetch\n",
+      "    return self.collate_fn(data)\n",
+      "  File \"<ipython-input-62-378242bfdb9f>\", line 7, in pad_batch_hyperband\n",
+      "    label_processed = LABEL.process(label)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 237, in process\n",
+      "    tensor = self.numericalize(padded, device=device)\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 338, in numericalize\n",
+      "    arr = [self.vocab.stoi[x] for x in arr]\n",
+      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 338, in <listcomp>\n",
+      "    arr = [self.vocab.stoi[x] for x in arr]\n",
+      "KeyError: tensor([0.])\n"
+     ]
+    },
+    {
+     "ename": "KeyError",
+     "evalue": "tensor([0.])",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-73-5415ff375924>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    671\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mdefault_client\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mif_delegate_has_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdelegate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_estimator_\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"estimator\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 347\u001b[0;31m         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    348\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    349\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mf\u001b[0;34m()\u001b[0m\n\u001b[1;32m    329\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mcallback_timeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    330\u001b[0m                 \u001b[0mfuture\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0masyncio\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait_for\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfuture\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 331\u001b[0;31m             \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    332\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    333\u001b[0m             \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_hyperband.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    398\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    399\u001b[0m         \u001b[0;31m# _fit is run in parallel because it's also a tornado coroutine\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 400\u001b[0;31m         \u001b[0m_SHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mSHAs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0m_brackets_ids\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    401\u001b[0m         \u001b[0mSHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_brackets_ids\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_SHAs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    402\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mcallback\u001b[0;34m(f)\u001b[0m\n\u001b[1;32m    847\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mchildren\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    848\u001b[0m                 \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 849\u001b[0;31m                     \u001b[0mresult_list\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    850\u001b[0m                 \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    851\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    623\u001b[0m             \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom_state\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    624\u001b[0m             \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 625\u001b[0;31m             \u001b[0mprefix\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprefix\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    626\u001b[0m         )\n\u001b[1;32m    627\u001b[0m         \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_process_results\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(model, params, X_train, y_train, X_test, y_test, additional_calls, fit_params, scorer, random_state, verbose, prefix)\u001b[0m\n\u001b[1;32m    231\u001b[0m     \u001b[0;31m# async for future, result in seq:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    232\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mitertools\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcount\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 233\u001b[0;31m         \u001b[0mmetas\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mclient\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnew_scores\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    235\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlog_delay\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlog_delay\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36m_gather\u001b[0;34m(self, futures, errors, direct, local_worker)\u001b[0m\n\u001b[1;32m   1824\u001b[0m                             \u001b[0mexc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCancelledError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1825\u001b[0m                         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1826\u001b[0;31m                             \u001b[0;32mraise\u001b[0m \u001b[0mexception\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1827\u001b[0m                         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1828\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0merrors\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m\"skip\"\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_score\u001b[0;34m()\u001b[0m\n\u001b[1;32m    101\u001b[0m     \u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmeta\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel_and_meta\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    102\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mscorer\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 103\u001b[0;31m         \u001b[0mscore\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscorer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    104\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    105\u001b[0m         \u001b[0mscore\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/sklearn/metrics/_scorer.py\u001b[0m in \u001b[0;36m_passthrough_scorer\u001b[0;34m()\u001b[0m\n\u001b[1;32m    369\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0m_passthrough_scorer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mestimator\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    370\u001b[0m     \u001b[0;34m\"\"\"Function that wraps estimator.score\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 371\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    372\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    373\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/sklearn/base.py\u001b[0m in \u001b[0;36mscore\u001b[0;34m()\u001b[0m\n\u001b[1;32m    367\u001b[0m         \"\"\"\n\u001b[1;32m    368\u001b[0m         \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m\u001b[0mmetrics\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0maccuracy_score\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 369\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0maccuracy_score\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    370\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    371\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/skorch/classifier.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m()\u001b[0m\n\u001b[1;32m    208\u001b[0m         \"\"\"\n\u001b[1;32m    209\u001b[0m         \u001b[0my_preds\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 210\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0myp\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward_iter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    211\u001b[0m             \u001b[0myp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0myp\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0myp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtuple\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0myp\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    212\u001b[0m             \u001b[0my_preds\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mto_numpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0myp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/skorch/net.py\u001b[0m in \u001b[0;36mforward_iter\u001b[0;34m()\u001b[0m\n\u001b[1;32m    915\u001b[0m         \u001b[0mdataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    916\u001b[0m         \u001b[0miterator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtraining\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 917\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0mdata\u001b[0m \u001b[0;32min\u001b[0m \u001b[0miterator\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    918\u001b[0m             \u001b[0mXi\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0munpack_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    919\u001b[0m             \u001b[0myp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mevaluation_step\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mXi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtraining\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m()\u001b[0m\n\u001b[1;32m    343\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    344\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__next__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 345\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    346\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_num_yielded\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    347\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_kind\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0m_DatasetKind\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mIterable\u001b[0m \u001b[0;32mand\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m_next_data\u001b[0;34m()\u001b[0m\n\u001b[1;32m    383\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    384\u001b[0m         \u001b[0mindex\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_index\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 385\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_fetcher\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfetch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    386\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_pin_memory\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    387\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_utils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\u001b[0m in \u001b[0;36mfetch\u001b[0;34m()\u001b[0m\n\u001b[1;32m     45\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     46\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mpossibly_batched_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 47\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcollate_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
+      "\u001b[0;32m<ipython-input-62-378242bfdb9f>\u001b[0m in \u001b[0;36mpad_batch_hyperband\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0;31m# numericalized and padded text representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0mtext_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTEXT\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtext\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m     \u001b[0mlabel_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mLABEL\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mtext_processed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabel_processed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mprocess\u001b[0;34m()\u001b[0m\n\u001b[1;32m    235\u001b[0m         \"\"\"\n\u001b[1;32m    236\u001b[0m         \u001b[0mpadded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 237\u001b[0;31m         \u001b[0mtensor\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumericalize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpadded\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    238\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mtensor\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    239\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mnumericalize\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mKeyError\u001b[0m: tensor([0.])"
      ]
     }
    ],
    "source": [
-    "# it's been erroring out with a less than helpful error message. This started happening right around \n",
-    "# when I started passing module__pretrained_embeddings=vocab.vectors. Unclear if this is the culprit.\n",
+    "# I think I'm getting yelled at for passing in text\n",
+    "# TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found <U13704\n",
     "\n",
     "# NB: this took ~3800 seconds on a single Nvidia GTX 980 Ti\n",
     "# notice how the number of training datapoints relates to the chunk size and our test_size\n",
@@ -1126,23 +1467,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "63.333333333333336"
-      ]
-     },
-     "execution_count": 50,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "3800 / 60"
-   ]
+   "outputs": [],
+   "source": []
   },
   {
    "cell_type": "markdown",
@@ -1160,56 +1488,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 45,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "0.8456"
-      ]
-     },
-     "execution_count": 45,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.best_score_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 46,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
-       "  module_=CNN(\n",
-       "    (embedding): Embedding(25002, 100)\n",
-       "    (conv_0): Conv1d(1, 50, kernel_size=(3, 100), stride=(1,))\n",
-       "    (conv_1): Conv1d(1, 50, kernel_size=(4, 100), stride=(1,))\n",
-       "    (conv_2): Conv1d(1, 50, kernel_size=(5, 100), stride=(1,))\n",
-       "    (fc): Linear(in_features=150, out_features=2, bias=True)\n",
-       "    (dropout): Dropout(p=0.22181138802671369, inplace=False)\n",
-       "  ),\n",
-       ")"
-      ]
-     },
-     "execution_count": 46,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.best_estimator_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 47,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1218,70 +1515,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 48,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "{'param_module__n_filters': array([50, 50, 50, 25, 50]),\n",
-       " 'test_score': array([0.8456, 0.7982, 0.733 , 0.843 , 0.8446]),\n",
-       " 'std_score_time': array([0.06216696, 0.19731486, 0.56393194, 1.33936262, 0.06445384]),\n",
-       " 'rank_test_score': array([1, 2, 3, 2, 1]),\n",
-       " 'mean_score_time': array([3.82462454, 4.0282222 , 5.47523713, 4.41082978, 5.29884744]),\n",
-       " 'params': array([{'batch_size': 64, 'module__dropout': 0.22181138802671369, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
-       "        {'batch_size': 64, 'module__dropout': 0.19188105885296006, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
-       "        {'batch_size': 32, 'module__dropout': 0.11893575283135104, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
-       "        {'batch_size': 32, 'module__dropout': 0.10900910968190201, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 25},\n",
-       "        {'batch_size': 32, 'module__dropout': 0.10009422737462631, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50}],\n",
-       "       dtype=object),\n",
-       " 'partial_fit_calls': array([6, 2, 2, 8, 8]),\n",
-       " 'param_module__dropout': array([0.22181139, 0.19188106, 0.11893575, 0.10900911, 0.10009423]),\n",
-       " 'std_partial_fit_time': array([ 3.6288025 ,  1.52035153, 11.09951091, 15.07569551,  5.98224938]),\n",
-       " 'mean_partial_fit_time': array([ 93.61417103,  96.36260712, 136.89576507, 143.69626164,\n",
-       "        130.64727747]),\n",
-       " 'model_id': array(['bracket=1-0', 'bracket=1-1', 'bracket=1-2', 'bracket=0-0',\n",
-       "        'bracket=0-1'], dtype='<U11'),\n",
-       " 'bracket': array([1, 1, 1, 0, 0]),\n",
-       " 'param_batch_size': array([64, 64, 32, 32, 32]),\n",
-       " 'param_module__filter_sizes': array([[3, 4, 5],\n",
-       "        [3, 4, 5],\n",
-       "        [3, 4, 5],\n",
-       "        [2, 3, 4],\n",
-       "        [3, 4, 5]])}"
-      ]
-     },
-     "execution_count": 48,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.cv_results_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "ename": "Exception",
-     "evalue": "Data must be 1-dimensional",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mException\u001b[0m                                 Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-49-245daea12296>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mcv_results\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDataFrame\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcv_results_\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mcv_results\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhead\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/frame.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, data, index, columns, dtype, copy)\u001b[0m\n\u001b[1;32m    409\u001b[0m             )\n\u001b[1;32m    410\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 411\u001b[0;31m             \u001b[0mmgr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0minit_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    412\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mma\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mMaskedArray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    413\u001b[0m             \u001b[0;32mimport\u001b[0m \u001b[0mnumpy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mma\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmrecords\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mmrecords\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36minit_dict\u001b[0;34m(data, index, columns, dtype)\u001b[0m\n\u001b[1;32m    255\u001b[0m             \u001b[0marr\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mis_datetime64tz_dtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marr\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0marr\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marrays\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    256\u001b[0m         ]\n\u001b[0;32m--> 257\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0marrays_to_mgr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marrays\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdata_names\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    258\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    259\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36marrays_to_mgr\u001b[0;34m(arrays, arr_names, index, columns, dtype)\u001b[0m\n\u001b[1;32m     80\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     81\u001b[0m     \u001b[0;31m# don't force copy because getting jammed in an ndarray anyway\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 82\u001b[0;31m     \u001b[0marrays\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_homogenize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0marrays\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     83\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     84\u001b[0m     \u001b[0;31m# from BlockManager perspective\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36m_homogenize\u001b[0;34m(data, index, dtype)\u001b[0m\n\u001b[1;32m    321\u001b[0m                 \u001b[0mval\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlib\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfast_multiget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moindex\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdefault\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnan\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    322\u001b[0m             val = sanitize_array(\n\u001b[0;32m--> 323\u001b[0;31m                 \u001b[0mval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mraise_cast_failure\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    324\u001b[0m             )\n\u001b[1;32m    325\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/pandas/core/internals/construction.py\u001b[0m in \u001b[0;36msanitize_array\u001b[0;34m(data, index, dtype, copy, raise_cast_failure)\u001b[0m\n\u001b[1;32m    727\u001b[0m     \u001b[0;32melif\u001b[0m \u001b[0msubarr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    728\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndarray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 729\u001b[0;31m             \u001b[0;32mraise\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Data must be 1-dimensional\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    730\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    731\u001b[0m             \u001b[0msubarr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masarray_tuplesafe\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;31mException\u001b[0m: Data must be 1-dimensional"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "cv_results = pd.DataFrame(search.cv_results_)\n",
     "cv_results.head()"
@@ -1289,7 +1534,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1299,178 +1544,9 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/html": [
-       "<div>\n",
-       "<style scoped>\n",
-       "    .dataframe tbody tr th:only-of-type {\n",
-       "        vertical-align: middle;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe tbody tr th {\n",
-       "        vertical-align: top;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe thead th {\n",
-       "        text-align: right;\n",
-       "    }\n",
-       "</style>\n",
-       "<table border=\"1\" class=\"dataframe\">\n",
-       "  <thead>\n",
-       "    <tr style=\"text-align: right;\">\n",
-       "      <th></th>\n",
-       "      <th>param_module__n_filters</th>\n",
-       "      <th>test_score</th>\n",
-       "      <th>std_score_time</th>\n",
-       "      <th>rank_test_score</th>\n",
-       "      <th>mean_score_time</th>\n",
-       "      <th>params</th>\n",
-       "      <th>partial_fit_calls</th>\n",
-       "      <th>param_module__dropout</th>\n",
-       "      <th>std_partial_fit_time</th>\n",
-       "      <th>mean_partial_fit_time</th>\n",
-       "      <th>model_id</th>\n",
-       "      <th>bracket</th>\n",
-       "      <th>param_batch_size</th>\n",
-       "      <th>param_module__filter_sizes</th>\n",
-       "    </tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr>\n",
-       "      <th>0</th>\n",
-       "      <td>50</td>\n",
-       "      <td>0.8456</td>\n",
-       "      <td>0.062167</td>\n",
-       "      <td>1</td>\n",
-       "      <td>3.824625</td>\n",
-       "      <td>{'batch_size': 64, 'module__dropout': 0.221811...</td>\n",
-       "      <td>6</td>\n",
-       "      <td>0.221811</td>\n",
-       "      <td>3.628802</td>\n",
-       "      <td>93.614171</td>\n",
-       "      <td>bracket=1-0</td>\n",
-       "      <td>1</td>\n",
-       "      <td>64</td>\n",
-       "      <td>[3, 4, 5]</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>1</th>\n",
-       "      <td>50</td>\n",
-       "      <td>0.7982</td>\n",
-       "      <td>0.197315</td>\n",
-       "      <td>2</td>\n",
-       "      <td>4.028222</td>\n",
-       "      <td>{'batch_size': 64, 'module__dropout': 0.191881...</td>\n",
-       "      <td>2</td>\n",
-       "      <td>0.191881</td>\n",
-       "      <td>1.520352</td>\n",
-       "      <td>96.362607</td>\n",
-       "      <td>bracket=1-1</td>\n",
-       "      <td>1</td>\n",
-       "      <td>64</td>\n",
-       "      <td>[3, 4, 5]</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>2</th>\n",
-       "      <td>50</td>\n",
-       "      <td>0.7330</td>\n",
-       "      <td>0.563932</td>\n",
-       "      <td>3</td>\n",
-       "      <td>5.475237</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.118935...</td>\n",
-       "      <td>2</td>\n",
-       "      <td>0.118936</td>\n",
-       "      <td>11.099511</td>\n",
-       "      <td>136.895765</td>\n",
-       "      <td>bracket=1-2</td>\n",
-       "      <td>1</td>\n",
-       "      <td>32</td>\n",
-       "      <td>[3, 4, 5]</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>3</th>\n",
-       "      <td>25</td>\n",
-       "      <td>0.8430</td>\n",
-       "      <td>1.339363</td>\n",
-       "      <td>2</td>\n",
-       "      <td>4.410830</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.109009...</td>\n",
-       "      <td>8</td>\n",
-       "      <td>0.109009</td>\n",
-       "      <td>15.075696</td>\n",
-       "      <td>143.696262</td>\n",
-       "      <td>bracket=0-0</td>\n",
-       "      <td>0</td>\n",
-       "      <td>32</td>\n",
-       "      <td>[2, 3, 4]</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>4</th>\n",
-       "      <td>50</td>\n",
-       "      <td>0.8446</td>\n",
-       "      <td>0.064454</td>\n",
-       "      <td>1</td>\n",
-       "      <td>5.298847</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.100094...</td>\n",
-       "      <td>8</td>\n",
-       "      <td>0.100094</td>\n",
-       "      <td>5.982249</td>\n",
-       "      <td>130.647277</td>\n",
-       "      <td>bracket=0-1</td>\n",
-       "      <td>0</td>\n",
-       "      <td>32</td>\n",
-       "      <td>[3, 4, 5]</td>\n",
-       "    </tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "</div>"
-      ],
-      "text/plain": [
-       "   param_module__n_filters  test_score  std_score_time  rank_test_score  \\\n",
-       "0                       50      0.8456        0.062167                1   \n",
-       "1                       50      0.7982        0.197315                2   \n",
-       "2                       50      0.7330        0.563932                3   \n",
-       "3                       25      0.8430        1.339363                2   \n",
-       "4                       50      0.8446        0.064454                1   \n",
-       "\n",
-       "   mean_score_time                                             params  \\\n",
-       "0         3.824625  {'batch_size': 64, 'module__dropout': 0.221811...   \n",
-       "1         4.028222  {'batch_size': 64, 'module__dropout': 0.191881...   \n",
-       "2         5.475237  {'batch_size': 32, 'module__dropout': 0.118935...   \n",
-       "3         4.410830  {'batch_size': 32, 'module__dropout': 0.109009...   \n",
-       "4         5.298847  {'batch_size': 32, 'module__dropout': 0.100094...   \n",
-       "\n",
-       "   partial_fit_calls  param_module__dropout  std_partial_fit_time  \\\n",
-       "0                  6               0.221811              3.628802   \n",
-       "1                  2               0.191881              1.520352   \n",
-       "2                  2               0.118936             11.099511   \n",
-       "3                  8               0.109009             15.075696   \n",
-       "4                  8               0.100094              5.982249   \n",
-       "\n",
-       "   mean_partial_fit_time     model_id  bracket  param_batch_size  \\\n",
-       "0              93.614171  bracket=1-0        1                64   \n",
-       "1              96.362607  bracket=1-1        1                64   \n",
-       "2             136.895765  bracket=1-2        1                32   \n",
-       "3             143.696262  bracket=0-0        0                32   \n",
-       "4             130.647277  bracket=0-1        0                32   \n",
-       "\n",
-       "  param_module__filter_sizes  \n",
-       "0                  [3, 4, 5]  \n",
-       "1                  [3, 4, 5]  \n",
-       "2                  [3, 4, 5]  \n",
-       "3                  [2, 3, 4]  \n",
-       "4                  [3, 4, 5]  "
-      ]
-     },
-     "execution_count": 61,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "cv_results = pd.DataFrame(search.cv_results_)\n",
     "cv_results.head()"
@@ -1478,98 +1554,27 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "0.83816"
-      ]
-     },
-     "execution_count": 62,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.score(X_test, y_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 63,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/html": [
-       "<table>\n",
-       "<tr>\n",
-       "<td>\n",
-       "<table>\n",
-       "  <thead>\n",
-       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr><th> Bytes </th><td> 200.00 kB </td> <td> 200.00 kB </td></tr>\n",
-       "    <tr><th> Shape </th><td> (25000,) </td> <td> (25000,) </td></tr>\n",
-       "    <tr><th> Count </th><td> 2 Tasks </td><td> 1 Chunks </td></tr>\n",
-       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "</td>\n",
-       "<td>\n",
-       "<svg width=\"170\" height=\"75\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
-       "\n",
-       "  <!-- Horizontal lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"120\" y2=\"0\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"0\" y1=\"25\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
-       "\n",
-       "  <!-- Vertical lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"25\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"120\" y1=\"0\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
-       "\n",
-       "  <!-- Colored Rectangle -->\n",
-       "  <polygon points=\"0.000000,0.000000 120.000000,0.000000 120.000000,25.412617 0.000000,25.412617\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
-       "\n",
-       "  <!-- Text -->\n",
-       "  <text x=\"60.000000\" y=\"45.412617\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >25000</text>\n",
-       "  <text x=\"140.000000\" y=\"12.706308\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(0,140.000000,12.706308)\">1</text>\n",
-       "</svg>\n",
-       "</td>\n",
-       "</tr>\n",
-       "</table>"
-      ],
-      "text/plain": [
-       "dask.array<_predict, shape=(25000,), dtype=int64, chunksize=(25000,), chunktype=numpy.ndarray>"
-      ]
-     },
-     "execution_count": 63,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.predict(X_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 64,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "array([0, 1, 1, ..., 0, 0, 0])"
-      ]
-     },
-     "execution_count": 64,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "search.predict(X_test).compute()"
    ]
@@ -1583,128 +1588,9 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 65,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/html": [
-       "<div>\n",
-       "<style scoped>\n",
-       "    .dataframe tbody tr th:only-of-type {\n",
-       "        vertical-align: middle;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe tbody tr th {\n",
-       "        vertical-align: top;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe thead th {\n",
-       "        text-align: right;\n",
-       "    }\n",
-       "</style>\n",
-       "<table border=\"1\" class=\"dataframe\">\n",
-       "  <thead>\n",
-       "    <tr style=\"text-align: right;\">\n",
-       "      <th></th>\n",
-       "      <th>model_id</th>\n",
-       "      <th>params</th>\n",
-       "      <th>partial_fit_calls</th>\n",
-       "      <th>partial_fit_time</th>\n",
-       "      <th>score</th>\n",
-       "      <th>score_time</th>\n",
-       "      <th>elapsed_wall_time</th>\n",
-       "      <th>bracket</th>\n",
-       "    </tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr>\n",
-       "      <th>0</th>\n",
-       "      <td>bracket=0-0</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.109009...</td>\n",
-       "      <td>1</td>\n",
-       "      <td>128.620566</td>\n",
-       "      <td>0.8272</td>\n",
-       "      <td>3.071467</td>\n",
-       "      <td>358.527012</td>\n",
-       "      <td>0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>1</th>\n",
-       "      <td>bracket=0-1</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.100094...</td>\n",
-       "      <td>1</td>\n",
-       "      <td>124.665028</td>\n",
-       "      <td>0.8386</td>\n",
-       "      <td>5.234394</td>\n",
-       "      <td>358.527018</td>\n",
-       "      <td>0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>2</th>\n",
-       "      <td>bracket=1-0</td>\n",
-       "      <td>{'batch_size': 64, 'module__dropout': 0.221811...</td>\n",
-       "      <td>1</td>\n",
-       "      <td>95.602639</td>\n",
-       "      <td>0.7558</td>\n",
-       "      <td>3.739039</td>\n",
-       "      <td>460.180206</td>\n",
-       "      <td>1</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>3</th>\n",
-       "      <td>bracket=1-1</td>\n",
-       "      <td>{'batch_size': 64, 'module__dropout': 0.191881...</td>\n",
-       "      <td>1</td>\n",
-       "      <td>94.842256</td>\n",
-       "      <td>0.7496</td>\n",
-       "      <td>4.225537</td>\n",
-       "      <td>460.180212</td>\n",
-       "      <td>1</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>4</th>\n",
-       "      <td>bracket=1-2</td>\n",
-       "      <td>{'batch_size': 32, 'module__dropout': 0.118935...</td>\n",
-       "      <td>1</td>\n",
-       "      <td>125.796254</td>\n",
-       "      <td>0.8306</td>\n",
-       "      <td>6.039169</td>\n",
-       "      <td>460.180215</td>\n",
-       "      <td>1</td>\n",
-       "    </tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "</div>"
-      ],
-      "text/plain": [
-       "      model_id                                             params  \\\n",
-       "0  bracket=0-0  {'batch_size': 32, 'module__dropout': 0.109009...   \n",
-       "1  bracket=0-1  {'batch_size': 32, 'module__dropout': 0.100094...   \n",
-       "2  bracket=1-0  {'batch_size': 64, 'module__dropout': 0.221811...   \n",
-       "3  bracket=1-1  {'batch_size': 64, 'module__dropout': 0.191881...   \n",
-       "4  bracket=1-2  {'batch_size': 32, 'module__dropout': 0.118935...   \n",
-       "\n",
-       "   partial_fit_calls  partial_fit_time   score  score_time  elapsed_wall_time  \\\n",
-       "0                  1        128.620566  0.8272    3.071467         358.527012   \n",
-       "1                  1        124.665028  0.8386    5.234394         358.527018   \n",
-       "2                  1         95.602639  0.7558    3.739039         460.180206   \n",
-       "3                  1         94.842256  0.7496    4.225537         460.180212   \n",
-       "4                  1        125.796254  0.8306    6.039169         460.180215   \n",
-       "\n",
-       "   bracket  \n",
-       "0        0  \n",
-       "1        0  \n",
-       "2        1  \n",
-       "3        1  \n",
-       "4        1  "
-      ]
-     },
-     "execution_count": 65,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "hist = pd.DataFrame(search.history_)\n",
     "hist.head()"

From 1e4ec4ba2547fd66e67dfb08b00eb04deb1b3e5d Mon Sep 17 00:00:00 2001
From: Todd Morrill <tmorrill12@gmail.com>
Date: Mon, 25 May 2020 11:40:49 -0400
Subject: [PATCH 5/6] working through hyperband errors

---
 machine-learning/skorch-hyperparam-opt.ipynb | 439 ++++++++++++-------
 1 file changed, 284 insertions(+), 155 deletions(-)

diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
index c06d1443..82544bf9 100644
--- a/machine-learning/skorch-hyperparam-opt.ipynb
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -37,7 +37,7 @@
        "<td style=\"vertical-align: top; border: 0px solid white\">\n",
        "<h3 style=\"text-align: left;\">Client</h3>\n",
        "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
-       "  <li><b>Scheduler: </b>tcp://127.0.0.1:42917</li>\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:38419</li>\n",
        "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
        "</ul>\n",
        "</td>\n",
@@ -53,7 +53,7 @@
        "</table>"
       ],
       "text/plain": [
-       "<Client: 'tcp://127.0.0.1:42917' processes=1 threads=1, memory=33.68 GB>"
+       "<Client: 'tcp://127.0.0.1:38419' processes=1 threads=1, memory=33.68 GB>"
       ]
      },
      "execution_count": 2,
@@ -83,15 +83,18 @@
    "outputs": [],
    "source": [
     "# for reproducibility\n",
-    "import random\n",
-    "import numpy as np\n",
+    "# NB: enabling reproducibility can significantly slow down runtimes\n",
+    "reproducible = True\n",
+    "if reproducible:\n",
+    "    import random\n",
+    "    import numpy as np\n",
     "\n",
-    "SEED = 42\n",
+    "    SEED = 42\n",
     "\n",
-    "random.seed(SEED)\n",
-    "np.random.seed(SEED)\n",
-    "torch.manual_seed(SEED)\n",
-    "torch.backends.cudnn.deterministic = True"
+    "    random.seed(SEED)\n",
+    "    np.random.seed(SEED)\n",
+    "    torch.manual_seed(SEED)\n",
+    "    torch.backends.cudnn.deterministic = True"
    ]
   },
   {
@@ -138,6 +141,13 @@
     "# make splits for data\n",
     "train, test = datasets.IMDB.splits(TEXT, LABEL)\n",
     "\n",
+    "# work with 5k datapoints for faster iteration times\n",
+    "split_ratio = 5_000 / len(train)\n",
+    "train, discard = train.split(split_ratio=split_ratio)\n",
+    "\n",
+    "split_ratio = 5_000 / len(test)\n",
+    "test, discard = test.split(split_ratio=split_ratio)\n",
+    "\n",
     "# will be used to initialize model embeddings layer\n",
     "vocab = torchtext.vocab.GloVe(name='6B', dim=100)\n",
     "\n",
@@ -214,7 +224,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.']\n",
+      "['it', 'aired', 'on', 'tv', 'yesterday,', 'so', 'i', 'decided', 'to', 'check', 'it', 'out.', 'this', 'was', 'one', 'of', 'the', 'last', 'bruce', 'timm/paul', 'dini', 'dtv', 'projects', 'related', 'to', 'their', 'old', '1992', 'batman', 'the', 'animated', 'series,', 'after', 'that', 'jeff', 'matsuda', 'came', 'along', 'and', 're-imagined', 'batman', 'with', 'his', 'new', 'the', 'batman', 'series,', 'but', 'anyway,', 'the', 'story', 'of', 'this', 'new', 'batman', 'movie', 'centers', 'around', 'the', 'appearance', 'of', 'a', 'new', 'vigilante', 'known', 'as', 'batwoman,', 'however', 'batman', 'feels', 'the', 'need', 'to', 'stop', 'her', 'because', 'of', 'her', 'extreme', 'methods,', 'and', 'also', 'in', 'the', 'meantime', 'take', 'down', 'the', 'pengiun', 'and', 'ruphert', 'thorn', 'who', 'both', 'are', 'secretly', 'working', 'with', 'carlton', \"duquesne(who's\", 'having', 'family', 'troubles)', 'and', 'another', 'villain(which', 'is', 'later', 'revealed', 'in', 'the', 'movie)', 'on', 'a', 'weapons', 'smuggling', 'operation,they', 'also', 'put', 'a', 'bounty', 'on', 'the', 'batwoman.', 'the', 'question', 'is:', 'who', 'is', 'this', 'mysterious', 'batwoman', 'and', 'is', 'it', 'possible', 'that', 'they', 'could', 'be', 'more', 'then', 'one?', \"it's\", 'up', 'to', 'batman', 'to', 'solve', 'this', 'mystery', 'and', 'stop', \"penguin's\", 'latest', 'operation.', 'for', 'an', 'animated', 'movie,', 'it', 'has', 'a', 'fairly', 'complex', 'plot', 'and', 'a', 'serious', 'tone,', 'which', 'is', 'good.', 'another', 'plus', 'was', 'the', 'complete', 'redesign', 'of', 'the', 'penguin', 'who', 'looks', 'much', 'more', 'like', 'the', 'sophisticated', 'mob', 'boss', \"we're\", 'used', 'to', 'seeing', 'in', 'the', 'comics,', 'unlike', 'his', 'previous', 'designs', 'that', 'borrowed', 'elements', 'from', 'tim', 'burton', 'vision', 'of', 'pengium(sewer', 'rat', 'and', 'circus', 'freak).', 'even', 'though', 'the', 'movie', 'contains', 'a', 'love', 'subplot', \"it's\", 'never', 'carried', 'that', 'far', 'and', \"doesn't\", 'derail', 'the', 'movie', 'like', 'say,', 'batman', 'forever.', 'the', 'voice', 'acting', 'is', 'standard', 'quality', 'for', 'these', 'direct-to-video', 'projects(if', 'only', 'batman:', 'mask', 'of', 'phantasm', 'took', 'this', 'route),', 'kevin', 'conroy', 'still', 'shines', 'as', 'batman/bruce', 'wayne.', 'and', 'like', 'i', 'said', 'despite', 'running', 'for', 'some', 'very', 'short', '80', 'minutes,', 'it', 'manages', 'to', 'make', 'a', 'pretty', 'good(and', 'complex)', 'storyline', 'complete', 'with', 'a', 'few', 'minor', 'twists', 'and', 'bucket', 'loads', 'of', 'action.', 'there', 'are', 'a', 'few', 'downsides,', 'however,', 'nightwing', 'is', 'nowhere', 'to', 'be', 'seen,', 'and', \"i'm\", 'sure', 'barbara', 'gordon', 'and', 'bruce', 'wayne', \"don't\", 'click', 'as', 'a', 'couple,', 'even', 'though', 'is', 'just', 'referenced,', 'tim', 'drake(aka', 'robin)', 'does', 'very', 'little', 'in', 'the', 'movie', 'and', 'to', 'be', 'quite', 'frank,', 'i', 'was', 'never', 'a', 'big', 'fan', 'of', 'paul', 'dini', 'and', 'bruce', \"timm's\", 'batman', 'character', 'design(especially', 'in', 'their', 'batman', 'shows', 'post-btas),', 'this', 'the', 'new', 'adventures', 'of', 'batman', 'and', 'robin,', 'well,', 'it', 'kinda', 'makes', 'batman', 'look', 'fat', 'then', 'rather', 'a', 'well-built', 'bulked', 'up', 'individual(kinda', 'like', 'the', 'jeff', 'matsuda', 'character', 'model', 'from', 'the', 'batman).', 'bruce', 'wayne', 'seems', 'a', 'bit', 'awkward,', 'those', 'blue', 'eyes', 'make', 'him', 'look', 'more', 'like', 'clark', 'kent', 'then', 'bruce(though', \"it's\", 'true', 'they', 'do', 'look', 'very', 'much', 'alike).', 'another', 'downside', 'is', 'rupert', 'throne(no', 'explanation', 'as', 'to', 'why', 'he', 'is', 'in', 'this', 'deal,', 'but', \"it's\", 'safe', 'to', 'say', \"he's\", 'has', 'goons', 'and', \"what's\", 'a', 'cut', 'of', 'the', 'deal)', 'which', 'does', 'very', 'little', 'more', 'then', 'hang', 'out', 'with', 'the', 'penguim', 'or', 'get', 'himself', 'hurt', 'every', 'time', 'he', 'points', 'a', 'gun', 'at', 'someone(count', 'how', 'many', 'times', 'this', 'happens', 'in', 'the', 'movie', 'and', \"you'll\", 'be', 'surprised.', 'overall,', 'a', 'good', 'batman', 'animated', 'movie,', 'worth', 'at', 'least', 'a', 'rental.']\n",
       "\n",
       "pos\n"
      ]
@@ -253,7 +263,7 @@
     {
      "data": {
       "text/plain": [
-       "torch.Size([2, 143])"
+       "torch.Size([2, 490])"
       ]
      },
      "execution_count": 12,
@@ -324,7 +334,7 @@
     {
      "data": {
       "text/plain": [
-       "'a masterful treatment of james caine\\'s \"the postman always rings twice\" as luchino visconti\\'s first film shot primarily around ferrara in a soulless war-torn italy. the original negative was thought destroyed but visconti saved a print and fortunately we can see this early neo-realist work today. a ruggedly handsome massimo girotti and clara calamai (who had recently revealed her breasts in la cena delle beffe\" (1941), star as the sensually-charged and ill-fated lovers who plot to kill her husband. unusual ending in which, although crime does not pay, one pays in a way not directly linked to the crime. excellent direction, script, acting, and cinematography. reportedly not as good as the french \"le dernier tournant\\' (1939) but probably better than the us version (1946) featuring lana turner and john garfield in the lead roles. highly recommended.'"
+       "\"it aired on tv yesterday, so i decided to check it out. this was one of the last bruce timm/paul dini dtv projects related to their old 1992 batman the animated series, after that jeff matsuda came along and re-imagined batman with his new the batman series, but anyway, the story of this new batman movie centers around the appearance of a new vigilante known as batwoman, however batman feels the need to stop her because of her extreme methods, and also in the meantime take down the pengiun and ruphert thorn who both are secretly working with carlton duquesne(who's having family troubles) and another villain(which is later revealed in the movie) on a weapons smuggling operation,they also put a bounty on the batwoman. the question is: who is this mysterious batwoman and is it possible that they could be more then one? it's up to batman to solve this mystery and stop penguin's latest operation. for an animated movie, it has a fairly complex plot and a serious tone, which is good. another plus was the complete redesign of the penguin who looks much more like the sophisticated mob boss we're used to seeing in the comics, unlike his previous designs that borrowed elements from tim burton vision of pengium(sewer rat and circus freak). even though the movie contains a love subplot it's never carried that far and doesn't derail the movie like say, batman forever. the voice acting is standard quality for these direct-to-video projects(if only batman: mask of phantasm took this route), kevin conroy still shines as batman/bruce wayne. and like i said despite running for some very short 80 minutes, it manages to make a pretty good(and complex) storyline complete with a few minor twists and bucket loads of action. there are a few downsides, however, nightwing is nowhere to be seen, and i'm sure barbara gordon and bruce wayne don't click as a couple, even though is just referenced, tim drake(aka robin) does very little in the movie and to be quite frank, i was never a big fan of paul dini and bruce timm's batman character design(especially in their batman shows post-btas), this the new adventures of batman and robin, well, it kinda makes batman look fat then rather a well-built bulked up individual(kinda like the jeff matsuda character model from the batman). bruce wayne seems a bit awkward, those blue eyes make him look more like clark kent then bruce(though it's true they do look very much alike). another downside is rupert throne(no explanation as to why he is in this deal, but it's safe to say he's has goons and what's a cut of the deal) which does very little more then hang out with the penguim or get himself hurt every time he points a gun at someone(count how many times this happens in the movie and you'll be surprised. overall, a good batman animated movie, worth at least a rental.\""
       ]
      },
      "execution_count": 16,
@@ -355,7 +365,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "(['a', 'masterful', 'treatment', 'of', 'james', \"caine's\", '\"the', 'postman', 'always', 'rings', 'twice\"', 'as', 'luchino', \"visconti's\", 'first', 'film', 'shot', 'primarily', 'around', 'ferrara', 'in', 'a', 'soulless', 'war-torn', 'italy.', 'the', 'original', 'negative', 'was', 'thought', 'destroyed', 'but', 'visconti', 'saved', 'a', 'print', 'and', 'fortunately', 'we', 'can', 'see', 'this', 'early', 'neo-realist', 'work', 'today.', 'a', 'ruggedly', 'handsome', 'massimo', 'girotti', 'and', 'clara', 'calamai', '(who', 'had', 'recently', 'revealed', 'her', 'breasts', 'in', 'la', 'cena', 'delle', 'beffe\"', '(1941),', 'star', 'as', 'the', 'sensually-charged', 'and', 'ill-fated', 'lovers', 'who', 'plot', 'to', 'kill', 'her', 'husband.', 'unusual', 'ending', 'in', 'which,', 'although', 'crime', 'does', 'not', 'pay,', 'one', 'pays', 'in', 'a', 'way', 'not', 'directly', 'linked', 'to', 'the', 'crime.', 'excellent', 'direction,', 'script,', 'acting,', 'and', 'cinematography.', 'reportedly', 'not', 'as', 'good', 'as', 'the', 'french', '\"le', 'dernier', \"tournant'\", '(1939)', 'but', 'probably', 'better', 'than', 'the', 'us', 'version', '(1946)', 'featuring', 'lana', 'turner', 'and', 'john', 'garfield', 'in', 'the', 'lead', 'roles.', 'highly', 'recommended.'], 'pos')\n"
+      "(['it', 'aired', 'on', 'tv', 'yesterday,', 'so', 'i', 'decided', 'to', 'check', 'it', 'out.', 'this', 'was', 'one', 'of', 'the', 'last', 'bruce', 'timm/paul', 'dini', 'dtv', 'projects', 'related', 'to', 'their', 'old', '1992', 'batman', 'the', 'animated', 'series,', 'after', 'that', 'jeff', 'matsuda', 'came', 'along', 'and', 're-imagined', 'batman', 'with', 'his', 'new', 'the', 'batman', 'series,', 'but', 'anyway,', 'the', 'story', 'of', 'this', 'new', 'batman', 'movie', 'centers', 'around', 'the', 'appearance', 'of', 'a', 'new', 'vigilante', 'known', 'as', 'batwoman,', 'however', 'batman', 'feels', 'the', 'need', 'to', 'stop', 'her', 'because', 'of', 'her', 'extreme', 'methods,', 'and', 'also', 'in', 'the', 'meantime', 'take', 'down', 'the', 'pengiun', 'and', 'ruphert', 'thorn', 'who', 'both', 'are', 'secretly', 'working', 'with', 'carlton', \"duquesne(who's\", 'having', 'family', 'troubles)', 'and', 'another', 'villain(which', 'is', 'later', 'revealed', 'in', 'the', 'movie)', 'on', 'a', 'weapons', 'smuggling', 'operation,they', 'also', 'put', 'a', 'bounty', 'on', 'the', 'batwoman.', 'the', 'question', 'is:', 'who', 'is', 'this', 'mysterious', 'batwoman', 'and', 'is', 'it', 'possible', 'that', 'they', 'could', 'be', 'more', 'then', 'one?', \"it's\", 'up', 'to', 'batman', 'to', 'solve', 'this', 'mystery', 'and', 'stop', \"penguin's\", 'latest', 'operation.', 'for', 'an', 'animated', 'movie,', 'it', 'has', 'a', 'fairly', 'complex', 'plot', 'and', 'a', 'serious', 'tone,', 'which', 'is', 'good.', 'another', 'plus', 'was', 'the', 'complete', 'redesign', 'of', 'the', 'penguin', 'who', 'looks', 'much', 'more', 'like', 'the', 'sophisticated', 'mob', 'boss', \"we're\", 'used', 'to', 'seeing', 'in', 'the', 'comics,', 'unlike', 'his', 'previous', 'designs', 'that', 'borrowed', 'elements', 'from', 'tim', 'burton', 'vision', 'of', 'pengium(sewer', 'rat', 'and', 'circus', 'freak).', 'even', 'though', 'the', 'movie', 'contains', 'a', 'love', 'subplot', \"it's\", 'never', 'carried', 'that', 'far', 'and', \"doesn't\", 'derail', 'the', 'movie', 'like', 'say,', 'batman', 'forever.', 'the', 'voice', 'acting', 'is', 'standard', 'quality', 'for', 'these', 'direct-to-video', 'projects(if', 'only', 'batman:', 'mask', 'of', 'phantasm', 'took', 'this', 'route),', 'kevin', 'conroy', 'still', 'shines', 'as', 'batman/bruce', 'wayne.', 'and', 'like', 'i', 'said', 'despite', 'running', 'for', 'some', 'very', 'short', '80', 'minutes,', 'it', 'manages', 'to', 'make', 'a', 'pretty', 'good(and', 'complex)', 'storyline', 'complete', 'with', 'a', 'few', 'minor', 'twists', 'and', 'bucket', 'loads', 'of', 'action.', 'there', 'are', 'a', 'few', 'downsides,', 'however,', 'nightwing', 'is', 'nowhere', 'to', 'be', 'seen,', 'and', \"i'm\", 'sure', 'barbara', 'gordon', 'and', 'bruce', 'wayne', \"don't\", 'click', 'as', 'a', 'couple,', 'even', 'though', 'is', 'just', 'referenced,', 'tim', 'drake(aka', 'robin)', 'does', 'very', 'little', 'in', 'the', 'movie', 'and', 'to', 'be', 'quite', 'frank,', 'i', 'was', 'never', 'a', 'big', 'fan', 'of', 'paul', 'dini', 'and', 'bruce', \"timm's\", 'batman', 'character', 'design(especially', 'in', 'their', 'batman', 'shows', 'post-btas),', 'this', 'the', 'new', 'adventures', 'of', 'batman', 'and', 'robin,', 'well,', 'it', 'kinda', 'makes', 'batman', 'look', 'fat', 'then', 'rather', 'a', 'well-built', 'bulked', 'up', 'individual(kinda', 'like', 'the', 'jeff', 'matsuda', 'character', 'model', 'from', 'the', 'batman).', 'bruce', 'wayne', 'seems', 'a', 'bit', 'awkward,', 'those', 'blue', 'eyes', 'make', 'him', 'look', 'more', 'like', 'clark', 'kent', 'then', 'bruce(though', \"it's\", 'true', 'they', 'do', 'look', 'very', 'much', 'alike).', 'another', 'downside', 'is', 'rupert', 'throne(no', 'explanation', 'as', 'to', 'why', 'he', 'is', 'in', 'this', 'deal,', 'but', \"it's\", 'safe', 'to', 'say', \"he's\", 'has', 'goons', 'and', \"what's\", 'a', 'cut', 'of', 'the', 'deal)', 'which', 'does', 'very', 'little', 'more', 'then', 'hang', 'out', 'with', 'the', 'penguim', 'or', 'get', 'himself', 'hurt', 'every', 'time', 'he', 'points', 'a', 'gun', 'at', 'someone(count', 'how', 'many', 'times', 'this', 'happens', 'in', 'the', 'movie', 'and', \"you'll\", 'be', 'surprised.', 'overall,', 'a', 'good', 'batman', 'animated', 'movie,', 'worth', 'at', 'least', 'a', 'rental.'], 'pos')\n"
      ]
     }
    ],
@@ -426,38 +436,38 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "tensor([[0.5732, 0.4268],\n",
-      "        [0.5372, 0.4628],\n",
-      "        [0.5615, 0.4385],\n",
-      "        [0.6806, 0.3194],\n",
-      "        [0.6352, 0.3648],\n",
-      "        [0.5678, 0.4322],\n",
-      "        [0.6505, 0.3495],\n",
-      "        [0.4325, 0.5675],\n",
-      "        [0.6019, 0.3981],\n",
-      "        [0.5857, 0.4143],\n",
-      "        [0.6010, 0.3990],\n",
-      "        [0.5662, 0.4338],\n",
-      "        [0.6494, 0.3506],\n",
-      "        [0.4390, 0.5610],\n",
-      "        [0.6041, 0.3959],\n",
-      "        [0.7121, 0.2879],\n",
-      "        [0.5781, 0.4219],\n",
-      "        [0.5727, 0.4273],\n",
-      "        [0.5169, 0.4831],\n",
-      "        [0.6531, 0.3469],\n",
-      "        [0.5080, 0.4920],\n",
-      "        [0.5611, 0.4389],\n",
-      "        [0.5560, 0.4440],\n",
-      "        [0.6448, 0.3552],\n",
-      "        [0.5697, 0.4303],\n",
-      "        [0.5798, 0.4202],\n",
-      "        [0.6656, 0.3344],\n",
-      "        [0.4933, 0.5067],\n",
-      "        [0.5143, 0.4857],\n",
-      "        [0.5518, 0.4482],\n",
-      "        [0.5806, 0.4194],\n",
-      "        [0.6070, 0.3930]], device='cuda:0', grad_fn=<SoftmaxBackward>)\n"
+      "tensor([[0.1738, 0.8262],\n",
+      "        [0.3462, 0.6538],\n",
+      "        [0.3317, 0.6683],\n",
+      "        [0.2430, 0.7570],\n",
+      "        [0.3659, 0.6341],\n",
+      "        [0.3846, 0.6154],\n",
+      "        [0.3757, 0.6243],\n",
+      "        [0.3713, 0.6287],\n",
+      "        [0.3153, 0.6847],\n",
+      "        [0.3656, 0.6344],\n",
+      "        [0.4096, 0.5904],\n",
+      "        [0.3088, 0.6912],\n",
+      "        [0.2832, 0.7168],\n",
+      "        [0.3461, 0.6539],\n",
+      "        [0.2674, 0.7326],\n",
+      "        [0.2894, 0.7106],\n",
+      "        [0.3850, 0.6150],\n",
+      "        [0.2820, 0.7180],\n",
+      "        [0.3537, 0.6463],\n",
+      "        [0.2389, 0.7611],\n",
+      "        [0.4709, 0.5291],\n",
+      "        [0.4199, 0.5801],\n",
+      "        [0.4443, 0.5557],\n",
+      "        [0.3591, 0.6409],\n",
+      "        [0.3270, 0.6730],\n",
+      "        [0.3097, 0.6903],\n",
+      "        [0.3241, 0.6759],\n",
+      "        [0.3594, 0.6406],\n",
+      "        [0.3175, 0.6825],\n",
+      "        [0.3447, 0.6553],\n",
+      "        [0.4566, 0.5434],\n",
+      "        [0.2943, 0.7057]], device='cuda:0', grad_fn=<SoftmaxBackward>)\n"
      ]
     }
    ],
@@ -574,8 +584,8 @@
      "text": [
       "  epoch    train_loss    valid_acc    valid_loss      dur\n",
       "-------  ------------  -----------  ------------  -------\n",
-      "      1        \u001b[36m0.4630\u001b[0m       \u001b[32m0.8312\u001b[0m        \u001b[35m0.3808\u001b[0m  33.8281\n",
-      "      2        \u001b[36m0.3454\u001b[0m       \u001b[32m0.8540\u001b[0m        \u001b[35m0.3361\u001b[0m  34.1240\n"
+      "      1        \u001b[36m0.6295\u001b[0m       \u001b[32m0.7840\u001b[0m        \u001b[35m0.5172\u001b[0m  12.0888\n",
+      "      2        \u001b[36m0.4530\u001b[0m       \u001b[32m0.8170\u001b[0m        \u001b[35m0.4310\u001b[0m  12.4245\n"
      ]
     },
     {
@@ -599,9 +609,6 @@
     }
    ],
    "source": [
-    "# NB: now that we implemented our custom batching we reduced compute time from 60 seconds\n",
-    "# per epoch to 20 seconds per epoch. That's huge.\n",
-    "# ((60 - 20) / 60)*100 = ~66%\n",
     "skorch_model.fit(train_dataset, y=None)"
    ]
   },
@@ -652,26 +659,6 @@
    "cell_type": "code",
    "execution_count": 35,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "array([0, 1, 1, ..., 0, 0, 0])"
-      ]
-     },
-     "execution_count": 35,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "test_preds"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 36,
-   "metadata": {},
    "outputs": [],
    "source": [
     "test_labels = processed_test_data[1].numpy()"
@@ -679,7 +666,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 37,
+   "execution_count": 36,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -688,16 +675,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 37,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "0.85048"
+       "0.8066"
       ]
      },
-     "execution_count": 38,
+     "execution_count": 37,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -708,16 +695,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 39,
+   "execution_count": 38,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "(array([0, 1]), array([12500, 12500]))"
+       "(array([0, 1]), array([2521, 2479]))"
       ]
      },
-     "execution_count": 39,
+     "execution_count": 38,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -729,7 +716,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 40,
+   "execution_count": 39,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -748,7 +735,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 40,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -758,7 +745,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 42,
+   "execution_count": 41,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -776,16 +763,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 43,
+   "execution_count": 42,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "['a masterful treatment of james caine\\'s \"the postman always rings twice\" as luchino visconti\\'s first film shot primarily around ferrara in a soulless war-torn italy. the original negative was thought destroyed but visconti saved a print and fortunately we can see this early neo-realist work today. a ruggedly handsome massimo girotti and clara calamai (who had recently revealed her breasts in la cena delle beffe\" (1941), star as the sensually-charged and ill-fated lovers who plot to kill her husband. unusual ending in which, although crime does not pay, one pays in a way not directly linked to the crime. excellent direction, script, acting, and cinematography. reportedly not as good as the french \"le dernier tournant\\' (1939) but probably better than the us version (1946) featuring lana turner and john garfield in the lead roles. highly recommended.']"
+       "[\"it aired on tv yesterday, so i decided to check it out. this was one of the last bruce timm/paul dini dtv projects related to their old 1992 batman the animated series, after that jeff matsuda came along and re-imagined batman with his new the batman series, but anyway, the story of this new batman movie centers around the appearance of a new vigilante known as batwoman, however batman feels the need to stop her because of her extreme methods, and also in the meantime take down the pengiun and ruphert thorn who both are secretly working with carlton duquesne(who's having family troubles) and another villain(which is later revealed in the movie) on a weapons smuggling operation,they also put a bounty on the batwoman. the question is: who is this mysterious batwoman and is it possible that they could be more then one? it's up to batman to solve this mystery and stop penguin's latest operation. for an animated movie, it has a fairly complex plot and a serious tone, which is good. another plus was the complete redesign of the penguin who looks much more like the sophisticated mob boss we're used to seeing in the comics, unlike his previous designs that borrowed elements from tim burton vision of pengium(sewer rat and circus freak). even though the movie contains a love subplot it's never carried that far and doesn't derail the movie like say, batman forever. the voice acting is standard quality for these direct-to-video projects(if only batman: mask of phantasm took this route), kevin conroy still shines as batman/bruce wayne. and like i said despite running for some very short 80 minutes, it manages to make a pretty good(and complex) storyline complete with a few minor twists and bucket loads of action. there are a few downsides, however, nightwing is nowhere to be seen, and i'm sure barbara gordon and bruce wayne don't click as a couple, even though is just referenced, tim drake(aka robin) does very little in the movie and to be quite frank, i was never a big fan of paul dini and bruce timm's batman character design(especially in their batman shows post-btas), this the new adventures of batman and robin, well, it kinda makes batman look fat then rather a well-built bulked up individual(kinda like the jeff matsuda character model from the batman). bruce wayne seems a bit awkward, those blue eyes make him look more like clark kent then bruce(though it's true they do look very much alike). another downside is rupert throne(no explanation as to why he is in this deal, but it's safe to say he's has goons and what's a cut of the deal) which does very little more then hang out with the penguim or get himself hurt every time he points a gun at someone(count how many times this happens in the movie and you'll be surprised. overall, a good batman animated movie, worth at least a rental.\"]"
       ]
      },
-     "execution_count": 43,
+     "execution_count": 42,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -796,7 +783,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 44,
+   "execution_count": 43,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -805,7 +792,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 45,
+   "execution_count": 44,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -814,7 +801,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 46,
+   "execution_count": 45,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -823,7 +810,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 47,
+   "execution_count": 46,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -832,17 +819,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 48,
+   "execution_count": 47,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "tensor([[    3,     0, 12500,  ...,     1,     1,     1],\n",
-       "        [ 1669,  3651,     9,  ...,  9300,  6517,   411]])"
+       "tensor([[    9, 12553,     0,  ...,     3,  7166,   369],\n",
+       "        [    3,  7166, 12553,  ...,     1,     1,     1]])"
       ]
      },
-     "execution_count": 48,
+     "execution_count": 47,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -853,16 +840,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": 48,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "torch.Size([2, 858])"
+       "torch.Size([2, 2825])"
       ]
      },
-     "execution_count": 49,
+     "execution_count": 48,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -873,7 +860,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": 49,
    "metadata": {},
    "outputs": [
     {
@@ -882,7 +869,7 @@
        "True"
       ]
      },
-     "execution_count": 50,
+     "execution_count": 49,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -893,7 +880,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 51,
+   "execution_count": 50,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -902,7 +889,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 52,
+   "execution_count": 51,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -911,7 +898,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 53,
+   "execution_count": 52,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -942,7 +929,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 54,
+   "execution_count": 53,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -952,12 +939,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 55,
+   "execution_count": 54,
    "metadata": {},
    "outputs": [],
    "source": [
     "# https://ml.dask.org/hyper-parameter-search.html#hyperband-parameters-rule-of-thumb\n",
-    "EPOCHS = 5\n",
+    "EPOCHS = 2\n",
     "NUM_TRAINING_EXAMPLES = len(train)*.8\n",
     "n_examples = EPOCHS * NUM_TRAINING_EXAMPLES\n",
     "n_params = 8\n",
@@ -969,7 +956,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 56,
+   "execution_count": 55,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -982,7 +969,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 57,
+   "execution_count": 56,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -991,16 +978,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 58,
+   "execution_count": 57,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Chunk size: 12500.0\n",
-      "Total chunks: 2\n",
-      "Last chunk size: 12500.0\n"
+      "Chunk size: 1000.0\n",
+      "Total chunks: 5\n",
+      "Last chunk size: 1000.0\n"
      ]
     }
    ],
@@ -1018,7 +1005,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 59,
+   "execution_count": 58,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1030,7 +1017,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": 59,
    "metadata": {},
    "outputs": [
     {
@@ -1044,10 +1031,10 @@
        "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
-       "    <tr><th> Bytes </th><td> 1.37 GB </td> <td> 685.20 MB </td></tr>\n",
-       "    <tr><th> Shape </th><td> (25000, 1) </td> <td> (12500, 1) </td></tr>\n",
-       "    <tr><th> Count </th><td> 3 Tasks </td><td> 2 Chunks </td></tr>\n",
-       "    <tr><th> Type </th><td> <U13704 </td><td> numpy.ndarray </td></tr>\n",
+       "    <tr><th> Bytes </th><td> 207.26 MB </td> <td> 41.45 MB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (5000, 1) </td> <td> (1000, 1) </td></tr>\n",
+       "    <tr><th> Count </th><td> 6 Tasks </td><td> 5 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> <U10363 </td><td> numpy.ndarray </td></tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</td>\n",
@@ -1056,7 +1043,10 @@
        "\n",
        "  <!-- Horizontal lines -->\n",
        "  <line x1=\"0\" y1=\"0\" x2=\"25\" y2=\"0\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"0\" y1=\"60\" x2=\"25\" y2=\"60\" />\n",
+       "  <line x1=\"0\" y1=\"24\" x2=\"25\" y2=\"24\" />\n",
+       "  <line x1=\"0\" y1=\"48\" x2=\"25\" y2=\"48\" />\n",
+       "  <line x1=\"0\" y1=\"72\" x2=\"25\" y2=\"72\" />\n",
+       "  <line x1=\"0\" y1=\"96\" x2=\"25\" y2=\"96\" />\n",
        "  <line x1=\"0\" y1=\"120\" x2=\"25\" y2=\"120\" style=\"stroke-width:2\" />\n",
        "\n",
        "  <!-- Vertical lines -->\n",
@@ -1068,17 +1058,17 @@
        "\n",
        "  <!-- Text -->\n",
        "  <text x=\"12.706308\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >1</text>\n",
-       "  <text x=\"45.412617\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,45.412617,60.000000)\">25000</text>\n",
+       "  <text x=\"45.412617\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,45.412617,60.000000)\">5000</text>\n",
        "</svg>\n",
        "</td>\n",
        "</tr>\n",
        "</table>"
       ],
       "text/plain": [
-       "dask.array<array, shape=(25000, 1), dtype=<U13704, chunksize=(12500, 1), chunktype=numpy.ndarray>"
+       "dask.array<array, shape=(5000, 1), dtype=<U10363, chunksize=(1000, 1), chunktype=numpy.ndarray>"
       ]
      },
-     "execution_count": 60,
+     "execution_count": 59,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1089,7 +1079,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 60,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1101,7 +1091,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": 61,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1119,7 +1109,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 63,
+   "execution_count": 62,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1150,7 +1140,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 64,
+   "execution_count": 63,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1159,7 +1149,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 65,
+   "execution_count": 64,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1173,7 +1163,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 66,
+   "execution_count": 65,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1182,7 +1172,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 67,
+   "execution_count": 66,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1197,7 +1187,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 68,
+   "execution_count": 67,
    "metadata": {},
    "outputs": [
     {
@@ -1206,7 +1196,7 @@
        "26"
       ]
      },
-     "execution_count": 68,
+     "execution_count": 67,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1217,7 +1207,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 69,
+   "execution_count": 68,
    "metadata": {},
    "outputs": [
     {
@@ -1226,7 +1216,7 @@
        "5"
       ]
      },
-     "execution_count": 69,
+     "execution_count": 68,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1237,7 +1227,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 70,
+   "execution_count": 69,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1246,7 +1236,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 71,
+   "execution_count": 70,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1261,7 +1251,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 72,
+   "execution_count": 71,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1273,28 +1263,34 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 83,
+   "execution_count": 72,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "'pos'"
-      ]
-     },
-     "execution_count": 83,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "y[0].compute()"
+    "# seems like the validation set is getting preprocessed twice\n",
+    "# KeyError: tensor([0.])\n",
+    "\n",
+    "# but then if you remove the collate_fn from iterator_valid__collate_fn=pad_batch_hyperband_partial\n",
+    "# then you get the following error\n",
+    "# TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found <U10363"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 73,
    "metadata": {},
+   "outputs": [],
+   "source": [
+    "# in general, I need to better understand what Hyperband is working on\n",
+    "# what happens after \"creating x models\"?\n",
+    "# it seems to take a fair amount of time before the GPU starts using power (i.e. model training)\n",
+    "# and then it's unclear how much progress has been made on the search"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 74,
+   "metadata": {},
    "outputs": [
     {
      "name": "stdout",
@@ -1319,8 +1315,8 @@
       "  sk_validation.check_array(sample, *args, **kwargs)\n",
       "/opt/conda/lib/python3.7/site-packages/torch/storage.py:34: FutureWarning: pickle support for Storage will be removed in 1.5. Use `torch.save` instead\n",
       "  warnings.warn(\"pickle support for Storage will be removed in 1.5. Use `torch.save` instead\", FutureWarning)\n",
-      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 23.79 MB detected in task graph: \n",
-      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .0721]]),\n",
+      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 21.54 MB detected in task graph: \n",
+      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .3232]]),\n",
       "), 0]\n",
       "Consider scattering large objects ahead of time\n",
       "with client.scatter to reduce scheduler burden and \n",
@@ -1358,8 +1354,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[CV, bracket=0] For training there are between 10000 and 10000 examples in each chunk\n",
-      "[CV, bracket=1] For training there are between 10000 and 10000 examples in each chunk\n"
+      "[CV, bracket=0] For training there are between 800 and 800 examples in each chunk\n",
+      "[CV, bracket=1] For training there are between 800 and 800 examples in each chunk\n"
      ]
     },
     {
@@ -1400,7 +1396,7 @@
       "    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration\n",
       "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\", line 47, in fetch\n",
       "    return self.collate_fn(data)\n",
-      "  File \"<ipython-input-62-378242bfdb9f>\", line 7, in pad_batch_hyperband\n",
+      "  File \"<ipython-input-61-378242bfdb9f>\", line 7, in pad_batch_hyperband\n",
       "    label_processed = LABEL.process(label)\n",
       "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 237, in process\n",
       "    tensor = self.numericalize(padded, device=device)\n",
@@ -1418,7 +1414,7 @@
      "traceback": [
       "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
       "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-73-5415ff375924>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m<ipython-input-74-5415ff375924>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    671\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mdefault_client\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mif_delegate_has_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdelegate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_estimator_\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"estimator\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 347\u001b[0;31m         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    348\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    349\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
@@ -1443,7 +1439,7 @@
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m()\u001b[0m\n\u001b[1;32m    343\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    344\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__next__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 345\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    346\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_num_yielded\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    347\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_kind\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0m_DatasetKind\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mIterable\u001b[0m \u001b[0;32mand\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m_next_data\u001b[0;34m()\u001b[0m\n\u001b[1;32m    383\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    384\u001b[0m         \u001b[0mindex\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_index\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 385\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_fetcher\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfetch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    386\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_pin_memory\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    387\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_utils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\u001b[0m in \u001b[0;36mfetch\u001b[0;34m()\u001b[0m\n\u001b[1;32m     45\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     46\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mpossibly_batched_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 47\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcollate_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
-      "\u001b[0;32m<ipython-input-62-378242bfdb9f>\u001b[0m in \u001b[0;36mpad_batch_hyperband\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0;31m# numericalized and padded text representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0mtext_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTEXT\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtext\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m     \u001b[0mlabel_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mLABEL\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mtext_processed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabel_processed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m<ipython-input-61-378242bfdb9f>\u001b[0m in \u001b[0;36mpad_batch_hyperband\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0;31m# numericalized and padded text representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0mtext_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTEXT\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtext\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m     \u001b[0mlabel_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mLABEL\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mtext_processed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabel_processed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mprocess\u001b[0;34m()\u001b[0m\n\u001b[1;32m    235\u001b[0m         \"\"\"\n\u001b[1;32m    236\u001b[0m         \u001b[0mpadded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 237\u001b[0;31m         \u001b[0mtensor\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumericalize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpadded\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    238\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mtensor\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    239\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mnumericalize\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
       "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
@@ -1467,10 +1463,143 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 75,
    "metadata": {},
    "outputs": [],
-   "source": []
+   "source": [
+    "# what's going on here?\n",
+    "# train_label should be text data and then get handled appropriately by pad_batch_hyperband\n",
+    "# instead, by the time the label makes it into pad_batch_hyperband, it is already a float tensor\n",
+    "# hence the lookup error. I was expecting a text string like 'pos'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 84,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array(['pos', 'pos'], dtype='<U3')"
+      ]
+     },
+     "execution_count": 84,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "train_label[:2]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 80,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<table>\n",
+       "<tr>\n",
+       "<td>\n",
+       "<table>\n",
+       "  <thead>\n",
+       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr><th> Bytes </th><td> 24 B </td> <td> 24 B </td></tr>\n",
+       "    <tr><th> Shape </th><td> (2,) </td> <td> (2,) </td></tr>\n",
+       "    <tr><th> Count </th><td> 7 Tasks </td><td> 1 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> <U3 </td><td> numpy.ndarray </td></tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</td>\n",
+       "<td>\n",
+       "<svg width=\"170\" height=\"110\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "\n",
+       "  <!-- Horizontal lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"120\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"60\" x2=\"120\" y2=\"60\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Vertical lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"60\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"120\" y1=\"0\" x2=\"120\" y2=\"60\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Colored Rectangle -->\n",
+       "  <polygon points=\"0.000000,0.000000 120.000000,0.000000 120.000000,60.000000 0.000000,60.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "\n",
+       "  <!-- Text -->\n",
+       "  <text x=\"60.000000\" y=\"80.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >2</text>\n",
+       "  <text x=\"140.000000\" y=\"30.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(0,140.000000,30.000000)\">1</text>\n",
+       "</svg>\n",
+       "</td>\n",
+       "</tr>\n",
+       "</table>"
+      ],
+      "text/plain": [
+       "dask.array<getitem, shape=(2,), dtype=<U3, chunksize=(2,), chunktype=numpy.ndarray>"
+      ]
+     },
+     "execution_count": 80,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "y[:2]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 82,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "dtype('<U3')"
+      ]
+     },
+     "execution_count": 82,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "y.dtype"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 81,
+   "metadata": {},
+   "outputs": [
+    {
+     "ename": "KeyboardInterrupt",
+     "evalue": "",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-81-61fb781fe4aa>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# why did this take so long and why is the GPU running?\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0my\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask/base.py\u001b[0m in \u001b[0;36mcompute\u001b[0;34m(self, **kwargs)\u001b[0m\n\u001b[1;32m    164\u001b[0m         \u001b[0mdask\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbase\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompute\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    165\u001b[0m         \"\"\"\n\u001b[0;32m--> 166\u001b[0;31m         \u001b[0;34m(\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraverse\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    167\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    168\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask/base.py\u001b[0m in \u001b[0;36mcompute\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    442\u001b[0m         \u001b[0mpostcomputes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__dask_postcompute__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    443\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 444\u001b[0;31m     \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mschedule\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdsk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkeys\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    445\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mrepack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpostcomputes\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    446\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36mget\u001b[0;34m(self, dsk, keys, restrictions, loose_restrictions, resources, sync, asynchronous, direct, retries, priority, fifo_timeout, actors, **kwargs)\u001b[0m\n\u001b[1;32m   2664\u001b[0m                     \u001b[0mshould_rejoin\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2665\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2666\u001b[0;31m                 \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpacked\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0masynchronous\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0masynchronous\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdirect\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdirect\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2667\u001b[0m             \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2668\u001b[0m                 \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mfutures\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36mgather\u001b[0;34m(self, futures, errors, direct, asynchronous)\u001b[0m\n\u001b[1;32m   1965\u001b[0m                 \u001b[0mdirect\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdirect\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1966\u001b[0m                 \u001b[0mlocal_worker\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mlocal_worker\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1967\u001b[0;31m                 \u001b[0masynchronous\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0masynchronous\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1968\u001b[0m             )\n\u001b[1;32m   1969\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    342\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    343\u001b[0m         \u001b[0;32mwhile\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mis_set\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 344\u001b[0;31m             \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    550\u001b[0m             \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flag\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    551\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 552\u001b[0;31m                 \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_cond\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    553\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    554\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;32m/opt/conda/lib/python3.7/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    298\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    299\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 300\u001b[0;31m                     \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    301\u001b[0m                 \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    302\u001b[0m                     \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
+     ]
+    }
+   ],
+   "source": [
+    "# why isn't this responding and why is the GPU running?\n",
+    "y[:2].compute()"
+   ]
   },
   {
    "cell_type": "markdown",

From 236381fe8416711f2264bbd56f97aa88996befd3 Mon Sep 17 00:00:00 2001
From: ToddMorrill <tmorrill12@gmail.com>
Date: Sat, 27 Jun 2020 16:58:48 +0000
Subject: [PATCH 6/6] working example

---
 machine-learning/skorch-hyperparam-opt.ipynb | 1613 +++++++++---------
 1 file changed, 807 insertions(+), 806 deletions(-)

diff --git a/machine-learning/skorch-hyperparam-opt.ipynb b/machine-learning/skorch-hyperparam-opt.ipynb
index 82544bf9..d551c689 100644
--- a/machine-learning/skorch-hyperparam-opt.ipynb
+++ b/machine-learning/skorch-hyperparam-opt.ipynb
@@ -6,8 +6,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install -q dask_cuda torch torchtext skorch\n",
-    "!pip -q install dask[dataframe] --upgrade"
+    "# !pip install -q dask_cuda torch torchtext skorch\n",
+    "# !pip -q install dask[dataframe] --upgrade"
    ]
   },
   {
@@ -21,13 +21,45 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Setup Dask"
+    "## Setup Dask Cluster"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 2,
    "metadata": {},
+   "outputs": [],
+   "source": [
+    "import math\n",
+    "import random\n",
+    "import time\n",
+    "\n",
+    "import dask.array as da\n",
+    "from dask_cuda import LocalCUDACluster\n",
+    "from dask_ml.model_selection import HyperbandSearchCV\n",
+    "from distributed import Client\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "from scipy.stats import loguniform\n",
+    "from sklearn.metrics import accuracy_score\n",
+    "from sklearn.model_selection import RandomizedSearchCV\n",
+    "import skorch\n",
+    "from skorch import NeuralNetClassifier\n",
+    "from skorch.helper import SliceDataset\n",
+    "import torch\n",
+    "import torch.nn as nn\n",
+    "import torch.nn.functional as F\n",
+    "import torch.optim as optim\n",
+    "from torch.utils.data import Dataset, DataLoader\n",
+    "import torchtext\n",
+    "from torchtext import data\n",
+    "from torchtext import datasets"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
    "outputs": [
     {
      "data": {
@@ -37,7 +69,7 @@
        "<td style=\"vertical-align: top; border: 0px solid white\">\n",
        "<h3 style=\"text-align: left;\">Client</h3>\n",
        "<ul style=\"text-align: left; list-style: none; margin: 0; padding: 0;\">\n",
-       "  <li><b>Scheduler: </b>tcp://127.0.0.1:38419</li>\n",
+       "  <li><b>Scheduler: </b>tcp://127.0.0.1:43999</li>\n",
        "  <li><b>Dashboard: </b><a href='http://127.0.0.1:8787/status' target='_blank'>http://127.0.0.1:8787/status</a></li>\n",
        "</ul>\n",
        "</td>\n",
@@ -46,26 +78,22 @@
        "<ul style=\"text-align: left; list-style:none; margin: 0; padding: 0;\">\n",
        "  <li><b>Workers: </b>1</li>\n",
        "  <li><b>Cores: </b>1</li>\n",
-       "  <li><b>Memory: </b>33.68 GB</li>\n",
+       "  <li><b>Memory: </b>31.63 GB</li>\n",
        "</ul>\n",
        "</td>\n",
        "</tr>\n",
        "</table>"
       ],
       "text/plain": [
-       "<Client: 'tcp://127.0.0.1:38419' processes=1 threads=1, memory=33.68 GB>"
+       "<Client: 'tcp://127.0.0.1:43999' processes=1 threads=1, memory=31.63 GB>"
       ]
      },
-     "execution_count": 2,
+     "execution_count": 3,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "import torch\n",
-    "from dask_cuda import LocalCUDACluster\n",
-    "from distributed import Client\n",
-    "\n",
     "# if you have GPU(s), use dask_cuda to automatically make use of them in your dask cluster\n",
     "if torch.cuda.is_available():\n",
     "    cluster = LocalCUDACluster()\n",
@@ -78,19 +106,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
     "# for reproducibility\n",
     "# NB: enabling reproducibility can significantly slow down runtimes\n",
-    "reproducible = True\n",
+    "reproducible = False\n",
     "if reproducible:\n",
-    "    import random\n",
-    "    import numpy as np\n",
-    "\n",
     "    SEED = 42\n",
-    "\n",
     "    random.seed(SEED)\n",
     "    np.random.seed(SEED)\n",
     "    torch.manual_seed(SEED)\n",
@@ -106,36 +130,35 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
-    "import torchtext\n",
-    "from torchtext import data\n",
-    "from torchtext import datasets\n",
-    "from torch.utils.data import Dataset, DataLoader\n",
-    "from torch.nn.utils.rnn import pad_sequence"
+    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
-    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
+    "# this solves many of our later problems but isn't an ideal solution\n",
+    "# accuracy will take a hit\n",
+    "FIX_LENGTH = 512"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# takes approx. 10 minutes to download data and embeddings (will be cached for re-use)\n",
+    "# a few seconds to download IMDB dataset (84Mb, will be cached)\n",
+    "# approx. 10 minutes to download glove embeddings (862Mb, will be cached)\n",
     "\n",
     "# set up fields\n",
-    "TEXT = data.Field(lower=True, batch_first=True, )\n",
+    "TEXT = data.Field(lower=True, batch_first=True, fix_length=FIX_LENGTH)\n",
     "LABEL = data.Field(sequential=False, unk_token=None)\n",
     "\n",
     "# make splits for data\n",
@@ -154,26 +177,21 @@
     "# build the vocabulary\n",
     "max_size = 25_000 # shorten for demonstrative purposes\n",
     "TEXT.build_vocab(train, vectors=vocab, max_size=max_size)\n",
-    "LABEL.build_vocab(train)\n",
-    "\n",
-    "# sadly I don't think we can use this \n",
-    "# make iterator for splits\n",
-    "# train_iter, test_iter = data.BucketIterator.splits(\n",
-    "#     (train, test), batch_sizes=(32, 64), device='cpu')"
+    "LABEL.build_vocab(train)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "['<unk>', '<pad>', 'the', 'a', 'and']"
+       "['<unk>', '<pad>', 'the', 'and', 'a']"
       ]
      },
-     "execution_count": 7,
+     "execution_count": 8,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -186,16 +204,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "defaultdict(None, {'neg': 0, 'pos': 1})"
+       "defaultdict(None, {'pos': 0, 'neg': 1})"
       ]
      },
-     "execution_count": 8,
+     "execution_count": 9,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -208,7 +226,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -217,14 +235,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
+   "execution_count": 11,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "['it', 'aired', 'on', 'tv', 'yesterday,', 'so', 'i', 'decided', 'to', 'check', 'it', 'out.', 'this', 'was', 'one', 'of', 'the', 'last', 'bruce', 'timm/paul', 'dini', 'dtv', 'projects', 'related', 'to', 'their', 'old', '1992', 'batman', 'the', 'animated', 'series,', 'after', 'that', 'jeff', 'matsuda', 'came', 'along', 'and', 're-imagined', 'batman', 'with', 'his', 'new', 'the', 'batman', 'series,', 'but', 'anyway,', 'the', 'story', 'of', 'this', 'new', 'batman', 'movie', 'centers', 'around', 'the', 'appearance', 'of', 'a', 'new', 'vigilante', 'known', 'as', 'batwoman,', 'however', 'batman', 'feels', 'the', 'need', 'to', 'stop', 'her', 'because', 'of', 'her', 'extreme', 'methods,', 'and', 'also', 'in', 'the', 'meantime', 'take', 'down', 'the', 'pengiun', 'and', 'ruphert', 'thorn', 'who', 'both', 'are', 'secretly', 'working', 'with', 'carlton', \"duquesne(who's\", 'having', 'family', 'troubles)', 'and', 'another', 'villain(which', 'is', 'later', 'revealed', 'in', 'the', 'movie)', 'on', 'a', 'weapons', 'smuggling', 'operation,they', 'also', 'put', 'a', 'bounty', 'on', 'the', 'batwoman.', 'the', 'question', 'is:', 'who', 'is', 'this', 'mysterious', 'batwoman', 'and', 'is', 'it', 'possible', 'that', 'they', 'could', 'be', 'more', 'then', 'one?', \"it's\", 'up', 'to', 'batman', 'to', 'solve', 'this', 'mystery', 'and', 'stop', \"penguin's\", 'latest', 'operation.', 'for', 'an', 'animated', 'movie,', 'it', 'has', 'a', 'fairly', 'complex', 'plot', 'and', 'a', 'serious', 'tone,', 'which', 'is', 'good.', 'another', 'plus', 'was', 'the', 'complete', 'redesign', 'of', 'the', 'penguin', 'who', 'looks', 'much', 'more', 'like', 'the', 'sophisticated', 'mob', 'boss', \"we're\", 'used', 'to', 'seeing', 'in', 'the', 'comics,', 'unlike', 'his', 'previous', 'designs', 'that', 'borrowed', 'elements', 'from', 'tim', 'burton', 'vision', 'of', 'pengium(sewer', 'rat', 'and', 'circus', 'freak).', 'even', 'though', 'the', 'movie', 'contains', 'a', 'love', 'subplot', \"it's\", 'never', 'carried', 'that', 'far', 'and', \"doesn't\", 'derail', 'the', 'movie', 'like', 'say,', 'batman', 'forever.', 'the', 'voice', 'acting', 'is', 'standard', 'quality', 'for', 'these', 'direct-to-video', 'projects(if', 'only', 'batman:', 'mask', 'of', 'phantasm', 'took', 'this', 'route),', 'kevin', 'conroy', 'still', 'shines', 'as', 'batman/bruce', 'wayne.', 'and', 'like', 'i', 'said', 'despite', 'running', 'for', 'some', 'very', 'short', '80', 'minutes,', 'it', 'manages', 'to', 'make', 'a', 'pretty', 'good(and', 'complex)', 'storyline', 'complete', 'with', 'a', 'few', 'minor', 'twists', 'and', 'bucket', 'loads', 'of', 'action.', 'there', 'are', 'a', 'few', 'downsides,', 'however,', 'nightwing', 'is', 'nowhere', 'to', 'be', 'seen,', 'and', \"i'm\", 'sure', 'barbara', 'gordon', 'and', 'bruce', 'wayne', \"don't\", 'click', 'as', 'a', 'couple,', 'even', 'though', 'is', 'just', 'referenced,', 'tim', 'drake(aka', 'robin)', 'does', 'very', 'little', 'in', 'the', 'movie', 'and', 'to', 'be', 'quite', 'frank,', 'i', 'was', 'never', 'a', 'big', 'fan', 'of', 'paul', 'dini', 'and', 'bruce', \"timm's\", 'batman', 'character', 'design(especially', 'in', 'their', 'batman', 'shows', 'post-btas),', 'this', 'the', 'new', 'adventures', 'of', 'batman', 'and', 'robin,', 'well,', 'it', 'kinda', 'makes', 'batman', 'look', 'fat', 'then', 'rather', 'a', 'well-built', 'bulked', 'up', 'individual(kinda', 'like', 'the', 'jeff', 'matsuda', 'character', 'model', 'from', 'the', 'batman).', 'bruce', 'wayne', 'seems', 'a', 'bit', 'awkward,', 'those', 'blue', 'eyes', 'make', 'him', 'look', 'more', 'like', 'clark', 'kent', 'then', 'bruce(though', \"it's\", 'true', 'they', 'do', 'look', 'very', 'much', 'alike).', 'another', 'downside', 'is', 'rupert', 'throne(no', 'explanation', 'as', 'to', 'why', 'he', 'is', 'in', 'this', 'deal,', 'but', \"it's\", 'safe', 'to', 'say', \"he's\", 'has', 'goons', 'and', \"what's\", 'a', 'cut', 'of', 'the', 'deal)', 'which', 'does', 'very', 'little', 'more', 'then', 'hang', 'out', 'with', 'the', 'penguim', 'or', 'get', 'himself', 'hurt', 'every', 'time', 'he', 'points', 'a', 'gun', 'at', 'someone(count', 'how', 'many', 'times', 'this', 'happens', 'in', 'the', 'movie', 'and', \"you'll\", 'be', 'surprised.', 'overall,', 'a', 'good', 'batman', 'animated', 'movie,', 'worth', 'at', 'least', 'a', 'rental.']\n",
+      "['i', \"couldn't\", 'hold', 'back', 'the', 'tears', 'when', 'i', 'watched', 'this'] ...\n",
       "\n",
       "pos\n"
      ]
@@ -232,53 +250,15 @@
    ],
    "source": [
     "# peek at the data\n",
-    "print(train.examples[0].text)\n",
+    "print(train.examples[0].text[:10], '...')\n",
     "print()\n",
     "print(train.examples[0].label)"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": 11,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# if you could use data.BucketIterator.splits\n",
-    "\n",
-    "# sadly train_iter isn't actually an iter..\n",
-    "# peek at a batch of data\n",
-    "# batch = next(iter(train_iter))\n",
-    "\n",
-    "# numericalized tokens\n",
-    "# print(batch.text)\n",
-    "\n",
-    "# print(batch.label)"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": 12,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "torch.Size([2, 490])"
-      ]
-     },
-     "execution_count": 12,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "TEXT.process([train[0].text, train[1].text]).shape"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 13,
-   "metadata": {},
    "outputs": [],
    "source": [
     "# custom dataset class required to work with Skorch\n",
@@ -296,89 +276,51 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 14,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# # custom dataset class required to work with Skorch\n",
-    "# class TorchDataset(torch.utils.data.Dataset):\n",
-    "#     def __init__(self, dataset, TEXT, LABEL):\n",
-    "#         self.dataset = dataset\n",
-    "#         self.TEXT = TEXT\n",
-    "#         self.LABEL = LABEL\n",
-    "    \n",
-    "#     def __getitem__(self, idx):\n",
-    "#         example = self.dataset.examples[idx]\n",
-    "#         X = TEXT.numericalize([example.text]).squeeze() # get rid of \"batch\" dimension\n",
-    "#         y = LABEL.numericalize([example.label])\n",
-    "#         return X, y\n",
-    "    \n",
-    "#     def __len__(self):\n",
-    "#         return len(self.dataset)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": 13,
    "metadata": {},
    "outputs": [],
    "source": [
-    "ex = train.examples[0]"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 16,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "\"it aired on tv yesterday, so i decided to check it out. this was one of the last bruce timm/paul dini dtv projects related to their old 1992 batman the animated series, after that jeff matsuda came along and re-imagined batman with his new the batman series, but anyway, the story of this new batman movie centers around the appearance of a new vigilante known as batwoman, however batman feels the need to stop her because of her extreme methods, and also in the meantime take down the pengiun and ruphert thorn who both are secretly working with carlton duquesne(who's having family troubles) and another villain(which is later revealed in the movie) on a weapons smuggling operation,they also put a bounty on the batwoman. the question is: who is this mysterious batwoman and is it possible that they could be more then one? it's up to batman to solve this mystery and stop penguin's latest operation. for an animated movie, it has a fairly complex plot and a serious tone, which is good. another plus was the complete redesign of the penguin who looks much more like the sophisticated mob boss we're used to seeing in the comics, unlike his previous designs that borrowed elements from tim burton vision of pengium(sewer rat and circus freak). even though the movie contains a love subplot it's never carried that far and doesn't derail the movie like say, batman forever. the voice acting is standard quality for these direct-to-video projects(if only batman: mask of phantasm took this route), kevin conroy still shines as batman/bruce wayne. and like i said despite running for some very short 80 minutes, it manages to make a pretty good(and complex) storyline complete with a few minor twists and bucket loads of action. there are a few downsides, however, nightwing is nowhere to be seen, and i'm sure barbara gordon and bruce wayne don't click as a couple, even though is just referenced, tim drake(aka robin) does very little in the movie and to be quite frank, i was never a big fan of paul dini and bruce timm's batman character design(especially in their batman shows post-btas), this the new adventures of batman and robin, well, it kinda makes batman look fat then rather a well-built bulked up individual(kinda like the jeff matsuda character model from the batman). bruce wayne seems a bit awkward, those blue eyes make him look more like clark kent then bruce(though it's true they do look very much alike). another downside is rupert throne(no explanation as to why he is in this deal, but it's safe to say he's has goons and what's a cut of the deal) which does very little more then hang out with the penguim or get himself hurt every time he points a gun at someone(count how many times this happens in the movie and you'll be surprised. overall, a good batman animated movie, worth at least a rental.\""
-      ]
-     },
-     "execution_count": 16,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "' '.join(ex.text)"
+    "train_dataset = TorchDataset(train)\n",
+    "test_dataset = TorchDataset(test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 14,
    "metadata": {},
    "outputs": [],
    "source": [
-    "train_dataset = TorchDataset(train)\n",
-    "test_dataset = TorchDataset(test)"
+    "tokens, label = train_dataset[0]"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "(['it', 'aired', 'on', 'tv', 'yesterday,', 'so', 'i', 'decided', 'to', 'check', 'it', 'out.', 'this', 'was', 'one', 'of', 'the', 'last', 'bruce', 'timm/paul', 'dini', 'dtv', 'projects', 'related', 'to', 'their', 'old', '1992', 'batman', 'the', 'animated', 'series,', 'after', 'that', 'jeff', 'matsuda', 'came', 'along', 'and', 're-imagined', 'batman', 'with', 'his', 'new', 'the', 'batman', 'series,', 'but', 'anyway,', 'the', 'story', 'of', 'this', 'new', 'batman', 'movie', 'centers', 'around', 'the', 'appearance', 'of', 'a', 'new', 'vigilante', 'known', 'as', 'batwoman,', 'however', 'batman', 'feels', 'the', 'need', 'to', 'stop', 'her', 'because', 'of', 'her', 'extreme', 'methods,', 'and', 'also', 'in', 'the', 'meantime', 'take', 'down', 'the', 'pengiun', 'and', 'ruphert', 'thorn', 'who', 'both', 'are', 'secretly', 'working', 'with', 'carlton', \"duquesne(who's\", 'having', 'family', 'troubles)', 'and', 'another', 'villain(which', 'is', 'later', 'revealed', 'in', 'the', 'movie)', 'on', 'a', 'weapons', 'smuggling', 'operation,they', 'also', 'put', 'a', 'bounty', 'on', 'the', 'batwoman.', 'the', 'question', 'is:', 'who', 'is', 'this', 'mysterious', 'batwoman', 'and', 'is', 'it', 'possible', 'that', 'they', 'could', 'be', 'more', 'then', 'one?', \"it's\", 'up', 'to', 'batman', 'to', 'solve', 'this', 'mystery', 'and', 'stop', \"penguin's\", 'latest', 'operation.', 'for', 'an', 'animated', 'movie,', 'it', 'has', 'a', 'fairly', 'complex', 'plot', 'and', 'a', 'serious', 'tone,', 'which', 'is', 'good.', 'another', 'plus', 'was', 'the', 'complete', 'redesign', 'of', 'the', 'penguin', 'who', 'looks', 'much', 'more', 'like', 'the', 'sophisticated', 'mob', 'boss', \"we're\", 'used', 'to', 'seeing', 'in', 'the', 'comics,', 'unlike', 'his', 'previous', 'designs', 'that', 'borrowed', 'elements', 'from', 'tim', 'burton', 'vision', 'of', 'pengium(sewer', 'rat', 'and', 'circus', 'freak).', 'even', 'though', 'the', 'movie', 'contains', 'a', 'love', 'subplot', \"it's\", 'never', 'carried', 'that', 'far', 'and', \"doesn't\", 'derail', 'the', 'movie', 'like', 'say,', 'batman', 'forever.', 'the', 'voice', 'acting', 'is', 'standard', 'quality', 'for', 'these', 'direct-to-video', 'projects(if', 'only', 'batman:', 'mask', 'of', 'phantasm', 'took', 'this', 'route),', 'kevin', 'conroy', 'still', 'shines', 'as', 'batman/bruce', 'wayne.', 'and', 'like', 'i', 'said', 'despite', 'running', 'for', 'some', 'very', 'short', '80', 'minutes,', 'it', 'manages', 'to', 'make', 'a', 'pretty', 'good(and', 'complex)', 'storyline', 'complete', 'with', 'a', 'few', 'minor', 'twists', 'and', 'bucket', 'loads', 'of', 'action.', 'there', 'are', 'a', 'few', 'downsides,', 'however,', 'nightwing', 'is', 'nowhere', 'to', 'be', 'seen,', 'and', \"i'm\", 'sure', 'barbara', 'gordon', 'and', 'bruce', 'wayne', \"don't\", 'click', 'as', 'a', 'couple,', 'even', 'though', 'is', 'just', 'referenced,', 'tim', 'drake(aka', 'robin)', 'does', 'very', 'little', 'in', 'the', 'movie', 'and', 'to', 'be', 'quite', 'frank,', 'i', 'was', 'never', 'a', 'big', 'fan', 'of', 'paul', 'dini', 'and', 'bruce', \"timm's\", 'batman', 'character', 'design(especially', 'in', 'their', 'batman', 'shows', 'post-btas),', 'this', 'the', 'new', 'adventures', 'of', 'batman', 'and', 'robin,', 'well,', 'it', 'kinda', 'makes', 'batman', 'look', 'fat', 'then', 'rather', 'a', 'well-built', 'bulked', 'up', 'individual(kinda', 'like', 'the', 'jeff', 'matsuda', 'character', 'model', 'from', 'the', 'batman).', 'bruce', 'wayne', 'seems', 'a', 'bit', 'awkward,', 'those', 'blue', 'eyes', 'make', 'him', 'look', 'more', 'like', 'clark', 'kent', 'then', 'bruce(though', \"it's\", 'true', 'they', 'do', 'look', 'very', 'much', 'alike).', 'another', 'downside', 'is', 'rupert', 'throne(no', 'explanation', 'as', 'to', 'why', 'he', 'is', 'in', 'this', 'deal,', 'but', \"it's\", 'safe', 'to', 'say', \"he's\", 'has', 'goons', 'and', \"what's\", 'a', 'cut', 'of', 'the', 'deal)', 'which', 'does', 'very', 'little', 'more', 'then', 'hang', 'out', 'with', 'the', 'penguim', 'or', 'get', 'himself', 'hurt', 'every', 'time', 'he', 'points', 'a', 'gun', 'at', 'someone(count', 'how', 'many', 'times', 'this', 'happens', 'in', 'the', 'movie', 'and', \"you'll\", 'be', 'surprised.', 'overall,', 'a', 'good', 'batman', 'animated', 'movie,', 'worth', 'at', 'least', 'a', 'rental.'], 'pos')\n"
+      "['i', \"couldn't\", 'hold', 'back', 'the', 'tears', 'when', 'i', 'watched', 'this'] ...\n",
+      "\n",
+      "pos\n"
      ]
     }
    ],
    "source": [
-    "print(train_dataset[0])"
+    "print(tokens[:10], '...')\n",
+    "print()\n",
+    "print(label)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 16,
    "metadata": {},
    "outputs": [],
    "source": [
+    "# custom collate function for DataLoader\n",
     "def pad_batch(batch, TEXT, LABEL):\n",
     "    text, label = list(zip(*batch))\n",
     "    # numericalized and padded text representation\n",
@@ -393,7 +335,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 17,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -402,13 +344,43 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
+   "execution_count": 18,
    "metadata": {},
    "outputs": [],
    "source": [
     "batch = next(iter(train_dataloader))"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "processed_examples, labels = batch"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor([   9,   20,    7, 3765,   23,    8,   54,  692,    2, 2384]) ...\n",
+      "\n",
+      "tensor(1)\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(processed_examples[0][:10], '...')\n",
+    "print()\n",
+    "print(labels[0])"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -418,56 +390,56 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# was having trouble with when model was defined in the notebook\n",
-    "# Can't get attribute ‘CNN' on <module ‘__main__'\n",
+    "# would have defined the class in this notebook but was getting the following error from Hyperband\n",
+    "# PicklingError: Can't pickle <class '__main__.CNN'>: attribute lookup CNN on __main__ failed\n",
     "from model import CNN"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "tensor([[0.1738, 0.8262],\n",
-      "        [0.3462, 0.6538],\n",
-      "        [0.3317, 0.6683],\n",
-      "        [0.2430, 0.7570],\n",
-      "        [0.3659, 0.6341],\n",
-      "        [0.3846, 0.6154],\n",
-      "        [0.3757, 0.6243],\n",
-      "        [0.3713, 0.6287],\n",
-      "        [0.3153, 0.6847],\n",
-      "        [0.3656, 0.6344],\n",
-      "        [0.4096, 0.5904],\n",
-      "        [0.3088, 0.6912],\n",
-      "        [0.2832, 0.7168],\n",
-      "        [0.3461, 0.6539],\n",
-      "        [0.2674, 0.7326],\n",
-      "        [0.2894, 0.7106],\n",
-      "        [0.3850, 0.6150],\n",
-      "        [0.2820, 0.7180],\n",
-      "        [0.3537, 0.6463],\n",
-      "        [0.2389, 0.7611],\n",
-      "        [0.4709, 0.5291],\n",
-      "        [0.4199, 0.5801],\n",
-      "        [0.4443, 0.5557],\n",
-      "        [0.3591, 0.6409],\n",
-      "        [0.3270, 0.6730],\n",
-      "        [0.3097, 0.6903],\n",
-      "        [0.3241, 0.6759],\n",
-      "        [0.3594, 0.6406],\n",
-      "        [0.3175, 0.6825],\n",
-      "        [0.3447, 0.6553],\n",
-      "        [0.4566, 0.5434],\n",
-      "        [0.2943, 0.7057]], device='cuda:0', grad_fn=<SoftmaxBackward>)\n"
+      "tensor([[0.5000, 0.5000],\n",
+      "        [0.5174, 0.4826],\n",
+      "        [0.4599, 0.5401],\n",
+      "        [0.3472, 0.6528],\n",
+      "        [0.3888, 0.6112],\n",
+      "        [0.4259, 0.5741],\n",
+      "        [0.3734, 0.6266],\n",
+      "        [0.3527, 0.6473],\n",
+      "        [0.4275, 0.5725],\n",
+      "        [0.4277, 0.5723],\n",
+      "        [0.5281, 0.4719],\n",
+      "        [0.4183, 0.5817],\n",
+      "        [0.4409, 0.5591],\n",
+      "        [0.4205, 0.5795],\n",
+      "        [0.4820, 0.5180],\n",
+      "        [0.3552, 0.6448],\n",
+      "        [0.3843, 0.6157],\n",
+      "        [0.3047, 0.6953],\n",
+      "        [0.5312, 0.4688],\n",
+      "        [0.4069, 0.5931],\n",
+      "        [0.3691, 0.6309],\n",
+      "        [0.3541, 0.6459],\n",
+      "        [0.2763, 0.7237],\n",
+      "        [0.4770, 0.5230],\n",
+      "        [0.3749, 0.6251],\n",
+      "        [0.4165, 0.5835],\n",
+      "        [0.4208, 0.5792],\n",
+      "        [0.5268, 0.4732],\n",
+      "        [0.4046, 0.5954],\n",
+      "        [0.5047, 0.4953],\n",
+      "        [0.3795, 0.6205],\n",
+      "        [0.4030, 0.5970]], device='cuda:0', grad_fn=<SoftmaxBackward>)\n"
      ]
     }
    ],
@@ -481,7 +453,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -490,7 +462,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -499,7 +471,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -510,32 +482,12 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Quick attempt at model training"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 27,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import skorch\n",
-    "from skorch import NeuralNetClassifier \n",
-    "import torch.optim as optim"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 28,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import torch.nn as nn"
+    "## Quick attempt at model training to debug any issues"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 26,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -575,17 +527,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 27,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "  epoch    train_loss    valid_acc    valid_loss      dur\n",
-      "-------  ------------  -----------  ------------  -------\n",
-      "      1        \u001b[36m0.6295\u001b[0m       \u001b[32m0.7840\u001b[0m        \u001b[35m0.5172\u001b[0m  12.0888\n",
-      "      2        \u001b[36m0.4530\u001b[0m       \u001b[32m0.8170\u001b[0m        \u001b[35m0.4310\u001b[0m  12.4245\n"
+      "  epoch    train_loss    valid_acc    valid_loss     dur\n",
+      "-------  ------------  -----------  ------------  ------\n",
+      "      1        \u001b[36m0.6258\u001b[0m       \u001b[32m0.7930\u001b[0m        \u001b[35m0.5002\u001b[0m  2.0929\n",
+      "      2        \u001b[36m0.4405\u001b[0m       \u001b[32m0.8250\u001b[0m        \u001b[35m0.3986\u001b[0m  1.9010\n"
      ]
     },
     {
@@ -603,7 +555,7 @@
        ")"
       ]
      },
-     "execution_count": 30,
+     "execution_count": 27,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -614,20 +566,31 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 28,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# this isn't working as expected\n",
+    "# https://github.com/skorch-dev/skorch/issues/641\n",
+    "\n",
     "# skorch_model.score(test_dataset)\n",
     "# TypeError: score() missing 1 required positional argument: 'y'\n",
     "# skorch_model.score(test_dataset, y=None)\n",
-    "# ValueError: Expected array-like (array or non-string sequence), got None"
+    "# ValueError: Expected array-like (array or non-string sequence), got None\n",
+    "\n",
+    "# can monkey patch skorch_model to achieve native scoring\n",
+    "# def score(self, X, y=None): \n",
+    "#     ds = self.get_dataset(X) \n",
+    "#     target_iterator = self.get_iterator(ds, training=False) \n",
+    "    \n",
+    "#     y_true = np.concatenate([skorch.utils.to_numpy(y) for _, y in target_iterator]) \n",
+    "#     y_pred = self.predict(X)\n",
+    "    \n",
+    "#     return accuracy_score(y_true, y_pred) "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 29,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -637,317 +600,217 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 33,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# sadly processing the entire test set twice just to score the model\n",
-    "processed_test_data = next(iter(test_dataloader))"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 34,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# quick check on the test set accuracy\n",
-    "test_preds = skorch_model.predict(test_dataset)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 35,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "test_labels = processed_test_data[1].numpy()"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 36,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from sklearn.metrics import accuracy_score"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 37,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [
     {
-     "data": {
-      "text/plain": [
-       "0.8066"
-      ]
-     },
-     "execution_count": 37,
-     "metadata": {},
-     "output_type": "execute_result"
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "0.8006\n"
+     ]
     }
    ],
    "source": [
-    "accuracy_score(test_labels, test_preds)"
+    "# test set accuracy\n",
+    "test_preds = skorch_model.predict(test_dataset)\n",
+    "processed_test_data = next(iter(test_dataloader))\n",
+    "test_labels = processed_test_data[1].numpy()\n",
+    "print(accuracy_score(test_labels, test_preds))"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "(array([0, 1]), array([2521, 2479]))"
+       "(array([0, 1]), array([2554, 2446]))"
       ]
      },
-     "execution_count": 38,
+     "execution_count": 31,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "# random guessing would 50% accuracy so the model is indeed training\n",
+    "# random guessing would 50% accuracy so the model is indeed training well\n",
     "np.unique(test_labels, return_counts=True)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 39,
+   "execution_count": 32,
    "metadata": {},
    "outputs": [],
    "source": [
     "# NB: this has no effect on GPU memory usage. If I keyboard interrupt, the workers get\n",
     "# restarted and memory usage goes down. Deleting these \"handler\" objects doesn't delete\n",
     "# GPU memory references on the workers. \n",
-    "del skorch_model"
+    "# del skorch_model"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Grid search with Hyperband"
+    "### Grid search with Skorch"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 40,
+   "execution_count": 33,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# try to store the raw text in dask arrays and handle preprocessing inside the network\n",
-    "# this isn't ideal because feature engineering really should be separate from modeling"
+    "# define parameter grid\n",
+    "params = {'module__filter_sizes': [(1, 2, 3), (2, 3, 4), (3, 4, 5)], \n",
+    "          'module__n_filters': [25, 50, 100],\n",
+    "          'module__dropout': loguniform(1e-1, 3e-1),\n",
+    "          'batch_size': [32, 64],\n",
+    "         }\n",
+    "\n",
+    "skorch_search = RandomizedSearchCV(skorch_model, params, n_iter=2, cv=5)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 34,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# raw text\n",
-    "train_text = []\n",
-    "train_label = []\n",
-    "for e in train.examples:\n",
-    "    tokenized_text = e.text\n",
-    "    text_string = ' '.join(tokenized_text)\n",
-    "    train_text.append(text_string)\n",
-    "    label = e.label\n",
-    "    # label = LABEL.process([e.label]).numpy()\n",
-    "    train_label.append(label)"
+    "# This errors out with: TypeError: fit() missing 1 required positional argument: 'y'\n",
+    "# skorch_search.fit(train_dataset, y=None)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 42,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "[\"it aired on tv yesterday, so i decided to check it out. this was one of the last bruce timm/paul dini dtv projects related to their old 1992 batman the animated series, after that jeff matsuda came along and re-imagined batman with his new the batman series, but anyway, the story of this new batman movie centers around the appearance of a new vigilante known as batwoman, however batman feels the need to stop her because of her extreme methods, and also in the meantime take down the pengiun and ruphert thorn who both are secretly working with carlton duquesne(who's having family troubles) and another villain(which is later revealed in the movie) on a weapons smuggling operation,they also put a bounty on the batwoman. the question is: who is this mysterious batwoman and is it possible that they could be more then one? it's up to batman to solve this mystery and stop penguin's latest operation. for an animated movie, it has a fairly complex plot and a serious tone, which is good. another plus was the complete redesign of the penguin who looks much more like the sophisticated mob boss we're used to seeing in the comics, unlike his previous designs that borrowed elements from tim burton vision of pengium(sewer rat and circus freak). even though the movie contains a love subplot it's never carried that far and doesn't derail the movie like say, batman forever. the voice acting is standard quality for these direct-to-video projects(if only batman: mask of phantasm took this route), kevin conroy still shines as batman/bruce wayne. and like i said despite running for some very short 80 minutes, it manages to make a pretty good(and complex) storyline complete with a few minor twists and bucket loads of action. there are a few downsides, however, nightwing is nowhere to be seen, and i'm sure barbara gordon and bruce wayne don't click as a couple, even though is just referenced, tim drake(aka robin) does very little in the movie and to be quite frank, i was never a big fan of paul dini and bruce timm's batman character design(especially in their batman shows post-btas), this the new adventures of batman and robin, well, it kinda makes batman look fat then rather a well-built bulked up individual(kinda like the jeff matsuda character model from the batman). bruce wayne seems a bit awkward, those blue eyes make him look more like clark kent then bruce(though it's true they do look very much alike). another downside is rupert throne(no explanation as to why he is in this deal, but it's safe to say he's has goons and what's a cut of the deal) which does very little more then hang out with the penguim or get himself hurt every time he points a gun at someone(count how many times this happens in the movie and you'll be surprised. overall, a good batman animated movie, worth at least a rental.\"]"
-      ]
-     },
-     "execution_count": 42,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "train_text[:1]"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 43,
+   "execution_count": 35,
    "metadata": {},
    "outputs": [],
    "source": [
-    "train_text = np.array(train_text)"
+    "# https://github.com/skorch-dev/skorch/issues/605#issuecomment-650580286"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 44,
+   "execution_count": 36,
    "metadata": {},
    "outputs": [],
    "source": [
-    "train_text = np.expand_dims(train_text, axis=-1)"
+    "# ValueError: Dataset does not have consistent lengths.\n",
+    "# dummy_y = np.zeros((len(train_dataset)))\n",
+    "# skorch_search.fit(train_dataset, y=dummy_y)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 45,
+   "execution_count": 37,
    "metadata": {},
    "outputs": [],
    "source": [
-    "sample_batch = [train_text[0], train_text[10]]"
+    "# ValueError: Dataset does not have consistent lengths.\n",
+    "# y = torch.cat([LABEL.process([pair[1]]) for pair in train_dataset]).numpy()\n",
+    "# skorch_search.fit(train_dataset, y=y)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 46,
+   "execution_count": 38,
    "metadata": {},
    "outputs": [],
    "source": [
-    "sample_text_batch = [i[0] for i in sample_batch]"
+    "# ValueError: Dataset does not have consistent lengths.\n",
+    "# skorch_search.fit(train_dataset, y=SliceDataset(train_dataset, idx=1))"
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 47,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "tensor([[    9, 12553,     0,  ...,     3,  7166,   369],\n",
-       "        [    3,  7166, 12553,  ...,     1,     1,     1]])"
-      ]
-     },
-     "execution_count": 47,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "TEXT.process(sample_text_batch)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 48,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "torch.Size([2, 2825])"
-      ]
-     },
-     "execution_count": 48,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
    "source": [
-    "TEXT.process([train_text[0][0], train_text[1][0]]).shape"
+    "## Grid search with Hyperband"
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 49,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "True"
-      ]
-     },
-     "execution_count": 49,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
    "source": [
-    "isinstance(train_text[:10], np.ndarray)"
+    "This is a really unfortunate hack to make deep learning batching semantics work with `Skorch` and `Dask`. The downside here is that we're no longer padding to the longest sequence in the batch, rather we're padding to the longest sequence in the *dataset*, which results in signifcantly more computation and thus significantly more time to train a model.\n",
+    "\n",
+    "Our solution was to set a max sequence length but that's not an ideal solution since you're still performing extra computation and accuracy does suffer."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": 39,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# TEXT.process(train_text[0])"
+    "# train=True shuffles the data\n",
+    "train_iter_skorch = torchtext.data.Iterator(train, batch_size=len(train), train=True, sort=False, device='cpu')\n",
+    "test_iter_skorch = torchtext.data.Iterator(test, batch_size=len(test), train=False, sort=False, device='cpu')"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 51,
+   "execution_count": 40,
    "metadata": {},
    "outputs": [],
    "source": [
-    "train_label = np.array(train_label)"
+    "# takes some time to numericalize the whole dataset\n",
+    "\n",
+    "# also notice that skorch and dask expect numpy arrays, which isn't ideal since it ties you to the cpu.\n",
+    "# meanwhile, projects like https://rapids.ai/ are moving toward all GPU computation, avoiding the cpu altogether.\n",
+    "for batch in train_iter_skorch:\n",
+    "    X_train = batch.text.numpy()\n",
+    "    y_train = batch.label.numpy()"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 52,
+   "execution_count": 41,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# # this is a really unfortunate hack to make torchtext batching semantics work with skorch and dask\n",
-    "# # the downside here is that we're no longer padding to the longest sequence in the batch, rather\n",
-    "# # we're padding to the longest sequence in the *dataset*, which results in signifcantly more\n",
-    "# # computation and thus significantly more time to train a model\n",
-    "# # of course, you could set a max sequence length but that's not an ideal solution\n",
-    "# # another solution would be to create a different dataset object, but then you can't use torchtext,\n",
-    "# # which really is quite handy\n",
-    "\n",
-    "# # train=True shuffles the data\n",
-    "# train_iter_skorch = torchtext.data.Iterator(train, batch_size=len(train), train=True, sort=False, device='cpu')\n",
-    "# test_iter_skorch = torchtext.data.Iterator(test, batch_size=len(test), train=False, sort=False, device='cpu')\n",
-    "\n",
-    "# # takes some time to numericalize the whole dataset\n",
-    "\n",
-    "# # also notice that skorch and dask expect numpy arrays, which isn't ideal since it ties you to the cpu.\n",
-    "# # meanwhile, projects like https://rapids.ai/ are moving toward all GPU computation, avoiding the cpu altogether.\n",
-    "# for batch in train_iter_skorch:\n",
-    "#     X_train = batch.text[0].numpy()\n",
-    "#     y_train = batch.label.numpy()\n",
-    "\n",
-    "# for batch in test_iter_skorch:\n",
-    "#     X_test = batch.text[0].numpy()\n",
-    "#     y_test = batch.label.numpy()"
+    "for batch in test_iter_skorch:\n",
+    "    X_test = batch.text.numpy()\n",
+    "    y_test = batch.label.numpy()"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 53,
+   "execution_count": 42,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(5000, 512)"
+      ]
+     },
+     "execution_count": 42,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "# notice how awfully large the second dimension is\n",
-    "# X_train.shape"
+    "X_train.shape"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 54,
+   "execution_count": 49,
    "metadata": {},
    "outputs": [],
    "source": [
     "# https://ml.dask.org/hyper-parameter-search.html#hyperband-parameters-rule-of-thumb\n",
-    "EPOCHS = 2\n",
+    "EPOCHS = 10\n",
     "NUM_TRAINING_EXAMPLES = len(train)*.8\n",
     "n_examples = EPOCHS * NUM_TRAINING_EXAMPLES\n",
-    "n_params = 8\n",
+    "n_params = 12\n",
     "\n",
     "# it's not immediately obvious to beginners how all these parameters interact with each other\n",
     "max_iter = n_params\n",
@@ -956,7 +819,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 55,
+   "execution_count": 50,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -969,25 +832,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 56,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import math"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 57,
+   "execution_count": 51,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Chunk size: 1000.0\n",
-      "Total chunks: 5\n",
-      "Last chunk size: 1000.0\n"
+      "Chunk size: 3333.0\n",
+      "Total chunks: 2\n",
+      "Last chunk size: 1667.0\n"
      ]
     }
    ],
@@ -1005,19 +859,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 58,
+   "execution_count": 52,
    "metadata": {},
    "outputs": [],
    "source": [
-    "import dask.array as da\n",
-    "\n",
-    "X = da.from_array(train_text, chunks=(chunk_size))\n",
-    "y = da.from_array(train_label, chunks=(chunk_size))"
+    "X = da.from_array(X_train, chunks=(chunk_size, X_train.shape[-1]))\n",
+    "y = da.from_array(y_train, chunks=(chunk_size))"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 59,
+   "execution_count": 53,
    "metadata": {},
    "outputs": [
     {
@@ -1031,44 +883,41 @@
        "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
-       "    <tr><th> Bytes </th><td> 207.26 MB </td> <td> 41.45 MB </td></tr>\n",
-       "    <tr><th> Shape </th><td> (5000, 1) </td> <td> (1000, 1) </td></tr>\n",
-       "    <tr><th> Count </th><td> 6 Tasks </td><td> 5 Chunks </td></tr>\n",
-       "    <tr><th> Type </th><td> <U10363 </td><td> numpy.ndarray </td></tr>\n",
+       "    <tr><th> Bytes </th><td> 20.48 MB </td> <td> 13.65 MB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (5000, 512) </td> <td> (3333, 512) </td></tr>\n",
+       "    <tr><th> Count </th><td> 3 Tasks </td><td> 2 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</td>\n",
        "<td>\n",
-       "<svg width=\"75\" height=\"170\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "<svg width=\"88\" height=\"170\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
        "\n",
        "  <!-- Horizontal lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"25\" y2=\"0\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"0\" y1=\"24\" x2=\"25\" y2=\"24\" />\n",
-       "  <line x1=\"0\" y1=\"48\" x2=\"25\" y2=\"48\" />\n",
-       "  <line x1=\"0\" y1=\"72\" x2=\"25\" y2=\"72\" />\n",
-       "  <line x1=\"0\" y1=\"96\" x2=\"25\" y2=\"96\" />\n",
-       "  <line x1=\"0\" y1=\"120\" x2=\"25\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"38\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"79\" x2=\"38\" y2=\"79\" />\n",
+       "  <line x1=\"0\" y1=\"120\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
        "\n",
        "  <!-- Vertical lines -->\n",
        "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"120\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"25\" y1=\"0\" x2=\"25\" y2=\"120\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"38\" y1=\"0\" x2=\"38\" y2=\"120\" style=\"stroke-width:2\" />\n",
        "\n",
        "  <!-- Colored Rectangle -->\n",
-       "  <polygon points=\"0.000000,0.000000 25.412617,0.000000 25.412617,120.000000 0.000000,120.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "  <polygon points=\"0.000000,0.000000 38.727884,0.000000 38.727884,120.000000 0.000000,120.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
        "\n",
        "  <!-- Text -->\n",
-       "  <text x=\"12.706308\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >1</text>\n",
-       "  <text x=\"45.412617\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,45.412617,60.000000)\">5000</text>\n",
+       "  <text x=\"19.363942\" y=\"140.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >512</text>\n",
+       "  <text x=\"58.727884\" y=\"60.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(-90,58.727884,60.000000)\">5000</text>\n",
        "</svg>\n",
        "</td>\n",
        "</tr>\n",
        "</table>"
       ],
       "text/plain": [
-       "dask.array<array, shape=(5000, 1), dtype=<U10363, chunksize=(1000, 1), chunktype=numpy.ndarray>"
+       "dask.array<array, shape=(5000, 512), dtype=int64, chunksize=(3333, 512), chunktype=numpy.ndarray>"
       ]
      },
-     "execution_count": 59,
+     "execution_count": 53,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1078,38 +927,35 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 60,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# need new collate_fn due to the following error\n",
-    "# ValueError: Expected 2D array, got 1D array instead:\n",
-    "# array=['1' '1' '1' ... '1' '1' '1'].\n",
-    "# Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample."
+    "TLDR; you can't use dask arrays with `torch.utils.data.Dataloader`, which means you have to do all your data preparation ahead of time"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 54,
    "metadata": {},
    "outputs": [],
    "source": [
-    "def pad_batch_hyperband(batch, TEXT, LABEL):\n",
-    "    text, label = list(zip(*batch))\n",
-    "    # unnecessary extra dimension\n",
-    "    text = [i[0] for i in text]\n",
-    "    # numericalized and padded text representation\n",
-    "    text_processed = TEXT.process(text)\n",
-    "    label_processed = LABEL.process(label)\n",
-    "    return text_processed, label_processed\n",
+    "# raw_train_dataset = [x for x in train_dataset]\n",
+    "# raw_train_dataset_array = np.array(raw_train_dataset, dtype=object)\n",
+    "# dask_dataset = da.from_array(raw_train_dataset_array, chunks=(chunk_size))\n",
+    "# dask_dataset[0].compute()\n",
     "\n",
-    "pad_batch_hyperband_partial = partial(pad_batch_hyperband, TEXT=TEXT, LABEL=LABEL)"
+    "# TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found <class 'dask.array.core.Array'>\n",
+    "# data_iter = DataLoader(dask_dataset)\n",
+    "# next(iter(data_iter))\n",
+    "\n",
+    "# # TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found object\n",
+    "# np_data_iter = DataLoader(raw_train_dataset_array)\n",
+    "# next(iter(np_data_iter))"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": 55,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1120,13 +966,7 @@
     "                lr=0.001,\n",
     "                optimizer=optim.Adam,\n",
     "                criterion=nn.NLLLoss,\n",
-    "                iterator_train=DataLoader,\n",
-    "                iterator_train__shuffle=True,\n",
     "                iterator_train__batch_size=32,\n",
-    "                iterator_train__collate_fn=pad_batch_hyperband_partial,\n",
-    "                iterator_valid=DataLoader,\n",
-    "                iterator_valid__collate_fn=pad_batch_hyperband_partial,\n",
-    "                iterator_valid__shuffle=False,\n",
     "                iterator_valid__batch_size=64,\n",
     "                train_split=None, # let hyperband handle it\n",
     "                module__n_filters=100,\n",
@@ -1140,16 +980,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 63,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from scipy.stats import loguniform"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 64,
+   "execution_count": 56,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1163,16 +994,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 65,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from dask_ml.model_selection import HyperbandSearchCV"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 66,
+   "execution_count": 57,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1187,16 +1009,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 67,
+   "execution_count": 58,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "26"
+       "85"
       ]
      },
-     "execution_count": 67,
+     "execution_count": 58,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1207,16 +1029,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 68,
+   "execution_count": 59,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "5"
+       "17"
       ]
      },
-     "execution_count": 68,
+     "execution_count": 59,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -1227,16 +1049,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 69,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import time"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 70,
+   "execution_count": 60,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1250,73 +1063,31 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 71,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# I feel like the GPU is spending a lot of time waiting for the CPU to preprocess data\n",
-    "# I've been watching GPU power consumption during training and it's not as high as during\n",
-    "# skorch training, so I tried to increase the number of workers the dataloader uses and\n",
-    "# got the following error: AssertionError: daemonic processes are not allowed to have children"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 72,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# seems like the validation set is getting preprocessed twice\n",
-    "# KeyError: tensor([0.])\n",
-    "\n",
-    "# but then if you remove the collate_fn from iterator_valid__collate_fn=pad_batch_hyperband_partial\n",
-    "# then you get the following error\n",
-    "# TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found <U10363"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 73,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# in general, I need to better understand what Hyperband is working on\n",
-    "# what happens after \"creating x models\"?\n",
-    "# it seems to take a fair amount of time before the GPU starts using power (i.e. model training)\n",
-    "# and then it's unclear how much progress has been made on the search"
+    "Running training on a Nvidia Tesla T4\\ "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 74,
+   "execution_count": 61,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[CV, bracket=1] creating 3 models\n"
+      "[CV, bracket=2] creating 9 models\n",
+      "[CV, bracket=1] creating 5 models\n"
      ]
     },
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/torch/storage.py:34: FutureWarning: pickle support for Storage will be removed in 1.5. Use `torch.save` instead\n",
-      "  warnings.warn(\"pickle support for Storage will be removed in 1.5. Use `torch.save` instead\", FutureWarning)\n",
-      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3379: UserWarning: Large object of size 21.54 MB detected in task graph: \n",
-      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .3232]]),\n",
+      "/opt/conda/lib/python3.7/site-packages/distributed/worker.py:3351: UserWarning: Large object of size 10.00 MB detected in task graph: \n",
+      "  [<class 'skorch.classifier.NeuralNetClassifier'>[u ... .0000]]),\n",
       "), 0]\n",
       "Consider scattering large objects ahead of time\n",
       "with client.scatter to reduce scheduler burden and \n",
@@ -1333,127 +1104,26 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[CV, bracket=0] creating 2 models\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/dask_ml/utils.py:175: FutureWarning: Beginning in version 0.22, arrays of bytes/strings will be converted to decimal numbers if dtype='numeric'. It is recommended that you convert the array to a float dtype before using it in scikit-learn, for example by using your_array = your_array.astype(np.float64).\n",
-      "  sk_validation.check_array(sample, *args, **kwargs)\n",
-      "/opt/conda/lib/python3.7/site-packages/torch/storage.py:34: FutureWarning: pickle support for Storage will be removed in 1.5. Use `torch.save` instead\n",
-      "  warnings.warn(\"pickle support for Storage will be removed in 1.5. Use `torch.save` instead\", FutureWarning)\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[CV, bracket=0] For training there are between 800 and 800 examples in each chunk\n",
-      "[CV, bracket=1] For training there are between 800 and 800 examples in each chunk\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "tornado.application - ERROR - Multiple exceptions in yield list\n",
-      "Traceback (most recent call last):\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 849, in callback\n",
-      "    result_list.append(f.result())\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
-      "    yielded = self.gen.throw(*exc_info)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 625, in _fit\n",
-      "    prefix=self.prefix,\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
-      "    value = future.result()\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1107, in run\n",
-      "    yielded = self.gen.throw(*exc_info)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 233, in _fit\n",
-      "    metas = yield client.gather(new_scores)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/tornado/gen.py\", line 1099, in run\n",
-      "    value = future.result()\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/distributed/client.py\", line 1826, in _gather\n",
-      "    raise exception.with_traceback(traceback)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\", line 103, in _score\n",
-      "    score = scorer(model, X, y)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/sklearn/metrics/_scorer.py\", line 371, in _passthrough_scorer\n",
-      "    return estimator.score(*args, **kwargs)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/sklearn/base.py\", line 369, in score\n",
-      "    return accuracy_score(y, self.predict(X), sample_weight=sample_weight)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/skorch/classifier.py\", line 210, in predict\n",
-      "    for yp in self.forward_iter(X, training=False):\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/skorch/net.py\", line 917, in forward_iter\n",
-      "    for data in iterator:\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\", line 345, in __next__\n",
-      "    data = self._next_data()\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\", line 385, in _next_data\n",
-      "    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\", line 47, in fetch\n",
-      "    return self.collate_fn(data)\n",
-      "  File \"<ipython-input-61-378242bfdb9f>\", line 7, in pad_batch_hyperband\n",
-      "    label_processed = LABEL.process(label)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 237, in process\n",
-      "    tensor = self.numericalize(padded, device=device)\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 338, in numericalize\n",
-      "    arr = [self.vocab.stoi[x] for x in arr]\n",
-      "  File \"/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\", line 338, in <listcomp>\n",
-      "    arr = [self.vocab.stoi[x] for x in arr]\n",
-      "KeyError: tensor([0.])\n"
-     ]
-    },
-    {
-     "ename": "KeyError",
-     "evalue": "tensor([0.])",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-74-5415ff375924>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mstart\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0msearch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    671\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mdefault_client\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mif_delegate_has_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdelegate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_estimator_\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"estimator\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 347\u001b[0;31m         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    348\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    349\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36mf\u001b[0;34m()\u001b[0m\n\u001b[1;32m    329\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mcallback_timeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    330\u001b[0m                 \u001b[0mfuture\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0masyncio\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait_for\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfuture\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 331\u001b[0;31m             \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    332\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    333\u001b[0m             \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_hyperband.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    398\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    399\u001b[0m         \u001b[0;31m# _fit is run in parallel because it's also a tornado coroutine\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 400\u001b[0;31m         \u001b[0m_SHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mSHAs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0m_brackets_ids\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    401\u001b[0m         \u001b[0mSHAs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mSHA\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_brackets_ids\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_SHAs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    402\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mcallback\u001b[0;34m(f)\u001b[0m\n\u001b[1;32m    847\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mchildren\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    848\u001b[0m                 \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 849\u001b[0;31m                     \u001b[0mresult_list\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    850\u001b[0m                 \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    851\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, y, **fit_params)\u001b[0m\n\u001b[1;32m    623\u001b[0m             \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom_state\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    624\u001b[0m             \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 625\u001b[0;31m             \u001b[0mprefix\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprefix\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    626\u001b[0m         )\n\u001b[1;32m    627\u001b[0m         \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_process_results\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1105\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0mexc_info\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1106\u001b[0m                         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1107\u001b[0;31m                             \u001b[0myielded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mthrow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1108\u001b[0m                         \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1109\u001b[0m                             \u001b[0;31m# Break up a reference to itself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(model, params, X_train, y_train, X_test, y_test, additional_calls, fit_params, scorer, random_state, verbose, prefix)\u001b[0m\n\u001b[1;32m    231\u001b[0m     \u001b[0;31m# async for future, result in seq:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    232\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mitertools\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcount\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 233\u001b[0;31m         \u001b[0mmetas\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32myield\u001b[0m \u001b[0mclient\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnew_scores\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    235\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlog_delay\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0m_i\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlog_delay\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/tornado/gen.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1097\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1098\u001b[0m                     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1099\u001b[0;31m                         \u001b[0mvalue\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfuture\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1100\u001b[0m                     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1101\u001b[0m                         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhad_exception\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36m_gather\u001b[0;34m(self, futures, errors, direct, local_worker)\u001b[0m\n\u001b[1;32m   1824\u001b[0m                             \u001b[0mexc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCancelledError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1825\u001b[0m                         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1826\u001b[0;31m                             \u001b[0;32mraise\u001b[0m \u001b[0mexception\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraceback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1827\u001b[0m                         \u001b[0;32mraise\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1828\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0merrors\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m\"skip\"\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask_ml/model_selection/_incremental.py\u001b[0m in \u001b[0;36m_score\u001b[0;34m()\u001b[0m\n\u001b[1;32m    101\u001b[0m     \u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmeta\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel_and_meta\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    102\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mscorer\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 103\u001b[0;31m         \u001b[0mscore\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscorer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    104\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    105\u001b[0m         \u001b[0mscore\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/sklearn/metrics/_scorer.py\u001b[0m in \u001b[0;36m_passthrough_scorer\u001b[0;34m()\u001b[0m\n\u001b[1;32m    369\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0m_passthrough_scorer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mestimator\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    370\u001b[0m     \u001b[0;34m\"\"\"Function that wraps estimator.score\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 371\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    372\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    373\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/sklearn/base.py\u001b[0m in \u001b[0;36mscore\u001b[0;34m()\u001b[0m\n\u001b[1;32m    367\u001b[0m         \"\"\"\n\u001b[1;32m    368\u001b[0m         \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m\u001b[0mmetrics\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0maccuracy_score\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 369\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0maccuracy_score\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    370\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    371\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/skorch/classifier.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m()\u001b[0m\n\u001b[1;32m    208\u001b[0m         \"\"\"\n\u001b[1;32m    209\u001b[0m         \u001b[0my_preds\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 210\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0myp\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward_iter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    211\u001b[0m             \u001b[0myp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0myp\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0myp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtuple\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0myp\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    212\u001b[0m             \u001b[0my_preds\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mto_numpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0myp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/skorch/net.py\u001b[0m in \u001b[0;36mforward_iter\u001b[0;34m()\u001b[0m\n\u001b[1;32m    915\u001b[0m         \u001b[0mdataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    916\u001b[0m         \u001b[0miterator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtraining\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 917\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0mdata\u001b[0m \u001b[0;32min\u001b[0m \u001b[0miterator\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    918\u001b[0m             \u001b[0mXi\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0munpack_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    919\u001b[0m             \u001b[0myp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mevaluation_step\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mXi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtraining\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m()\u001b[0m\n\u001b[1;32m    343\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    344\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__next__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 345\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    346\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_num_yielded\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    347\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_kind\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0m_DatasetKind\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mIterable\u001b[0m \u001b[0;32mand\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/dataloader.py\u001b[0m in \u001b[0;36m_next_data\u001b[0;34m()\u001b[0m\n\u001b[1;32m    383\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_next_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    384\u001b[0m         \u001b[0mindex\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_index\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 385\u001b[0;31m         \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset_fetcher\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfetch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# may raise StopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    386\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_pin_memory\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    387\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_utils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpin_memory\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py\u001b[0m in \u001b[0;36mfetch\u001b[0;34m()\u001b[0m\n\u001b[1;32m     45\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     46\u001b[0m             \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mpossibly_batched_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 47\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcollate_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
-      "\u001b[0;32m<ipython-input-61-378242bfdb9f>\u001b[0m in \u001b[0;36mpad_batch_hyperband\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0;31m# numericalized and padded text representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m     \u001b[0mtext_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTEXT\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtext\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m     \u001b[0mlabel_processed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mLABEL\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprocess\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mtext_processed\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabel_processed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mprocess\u001b[0;34m()\u001b[0m\n\u001b[1;32m    235\u001b[0m         \"\"\"\n\u001b[1;32m    236\u001b[0m         \u001b[0mpadded\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 237\u001b[0;31m         \u001b[0mtensor\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumericalize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpadded\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    238\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mtensor\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    239\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36mnumericalize\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/torchtext/data/field.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m()\u001b[0m\n\u001b[1;32m    336\u001b[0m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mex\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mex\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    337\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 338\u001b[0;31m                 \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstoi\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    339\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    340\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpostprocessing\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;31mKeyError\u001b[0m: tensor([0.])"
+      "[CV, bracket=0] creating 3 models\n",
+      "[CV, bracket=0] For training there are between 1333 and 2666 examples in each chunk\n",
+      "[CV, bracket=1] For training there are between 1333 and 2666 examples in each chunk\n",
+      "[CV, bracket=2] For training there are between 1333 and 2666 examples in each chunk\n",
+      "[CV, bracket=0] validation score of 0.7982 received after 1 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8032 received after 1 partial_fit calls\n",
+      "[CV, bracket=2] validation score of 0.7842 received after 1 partial_fit calls\n",
+      "[CV, bracket=0] validation score of 0.8551 received after 12 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8322 received after 4 partial_fit calls\n",
+      "[CV, bracket=2] validation score of 0.8212 received after 3 partial_fit calls\n",
+      "[CV, bracket=1] validation score of 0.8162 received after 12 partial_fit calls\n",
+      "[CV, bracket=2] validation score of 0.8062 received after 9 partial_fit calls\n",
+      "Time to complete grid search: 372.02 seconds\n"
      ]
     }
    ],
    "source": [
-    "# I think I'm getting yelled at for passing in text\n",
-    "# TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found <U13704\n",
-    "\n",
-    "# NB: this took ~3800 seconds on a single Nvidia GTX 980 Ti\n",
     "# notice how the number of training datapoints relates to the chunk size and our test_size\n",
-    "# e.g. Validation set size: 2500 = 12500*.2, Train set size: 10000 = 12500 - 2500\n",
+    "# Train set chunk size: 800 = 1000*(1-.2)\n",
+    "# Validation set chunk size: 200 = 1000*.2\n",
     "start = time.time()\n",
     "search.fit(X, y)\n",
     "end = time.time()\n",
@@ -1462,208 +1132,182 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 75,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# what's going on here?\n",
-    "# train_label should be text data and then get handled appropriately by pad_batch_hyperband\n",
-    "# instead, by the time the label makes it into pad_batch_hyperband, it is already a float tensor\n",
-    "# hence the lookup error. I was expecting a text string like 'pos'"
+    "## Integration"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "`HyperbandSearchCV` follows the Scikit-learn API and mirrors Scikit-learn's `RandomizedSearchCV`. This means that it \"just works\". All the Scikit-learn attributes and methods are available:"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 84,
+   "execution_count": 62,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "array(['pos', 'pos'], dtype='<U3')"
+       "0.8551448551448552"
       ]
      },
-     "execution_count": 84,
+     "execution_count": 62,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "train_label[:2]"
+    "search.best_score_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 80,
+   "execution_count": 63,
    "metadata": {},
    "outputs": [
     {
      "data": {
-      "text/html": [
-       "<table>\n",
-       "<tr>\n",
-       "<td>\n",
-       "<table>\n",
-       "  <thead>\n",
-       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr><th> Bytes </th><td> 24 B </td> <td> 24 B </td></tr>\n",
-       "    <tr><th> Shape </th><td> (2,) </td> <td> (2,) </td></tr>\n",
-       "    <tr><th> Count </th><td> 7 Tasks </td><td> 1 Chunks </td></tr>\n",
-       "    <tr><th> Type </th><td> <U3 </td><td> numpy.ndarray </td></tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "</td>\n",
-       "<td>\n",
-       "<svg width=\"170\" height=\"110\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
-       "\n",
-       "  <!-- Horizontal lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"120\" y2=\"0\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"0\" y1=\"60\" x2=\"120\" y2=\"60\" style=\"stroke-width:2\" />\n",
-       "\n",
-       "  <!-- Vertical lines -->\n",
-       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"60\" style=\"stroke-width:2\" />\n",
-       "  <line x1=\"120\" y1=\"0\" x2=\"120\" y2=\"60\" style=\"stroke-width:2\" />\n",
-       "\n",
-       "  <!-- Colored Rectangle -->\n",
-       "  <polygon points=\"0.000000,0.000000 120.000000,0.000000 120.000000,60.000000 0.000000,60.000000\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
-       "\n",
-       "  <!-- Text -->\n",
-       "  <text x=\"60.000000\" y=\"80.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >2</text>\n",
-       "  <text x=\"140.000000\" y=\"30.000000\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(0,140.000000,30.000000)\">1</text>\n",
-       "</svg>\n",
-       "</td>\n",
-       "</tr>\n",
-       "</table>"
-      ],
       "text/plain": [
-       "dask.array<getitem, shape=(2,), dtype=<U3, chunksize=(2,), chunktype=numpy.ndarray>"
+       "<class 'skorch.classifier.NeuralNetClassifier'>[initialized](\n",
+       "  module_=CNN(\n",
+       "    (embedding): Embedding(25002, 100)\n",
+       "    (conv_0): Conv1d(1, 50, kernel_size=(1, 100), stride=(1,))\n",
+       "    (conv_1): Conv1d(1, 50, kernel_size=(2, 100), stride=(1,))\n",
+       "    (conv_2): Conv1d(1, 50, kernel_size=(3, 100), stride=(1,))\n",
+       "    (fc): Linear(in_features=150, out_features=2, bias=True)\n",
+       "    (dropout): Dropout(p=0.12476236679704862, inplace=False)\n",
+       "  ),\n",
+       ")"
       ]
      },
-     "execution_count": 80,
+     "execution_count": 63,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "y[:2]"
+    "search.best_estimator_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 82,
+   "execution_count": 64,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "dtype('<U3')"
+       "{'param_module__n_filters': array([100, 100,  25,  50,  25,  25,  25,  50,  50,  25, 100,  25, 100,\n",
+       "        100,  50, 100, 100]),\n",
+       " 'param_batch_size': array([32, 64, 64, 32, 32, 32, 32, 32, 32, 32, 64, 64, 64, 32, 32, 64, 32]),\n",
+       " 'param_module__dropout': array([0.14233276, 0.15454961, 0.273504  , 0.26444226, 0.13400896,\n",
+       "        0.19302409, 0.15142034, 0.23147478, 0.28516434, 0.15385363,\n",
+       "        0.18147304, 0.29957988, 0.10990513, 0.29003914, 0.12476237,\n",
+       "        0.14516028, 0.10459089]),\n",
+       " 'std_partial_fit_time': array([0.        , 1.42792781, 0.        , 0.        , 0.8889792 ,\n",
+       "        0.94623792, 0.        , 0.        , 0.        , 0.84567356,\n",
+       "        1.6097213 , 0.79452193, 1.30084145, 1.54377656, 0.04432201,\n",
+       "        0.26346576, 0.31667006]),\n",
+       " 'param_module__filter_sizes': array([[3, 4, 5],\n",
+       "        [2, 3, 4],\n",
+       "        [1, 2, 3],\n",
+       "        [2, 3, 4],\n",
+       "        [2, 3, 4],\n",
+       "        [3, 4, 5],\n",
+       "        [1, 2, 3],\n",
+       "        [3, 4, 5],\n",
+       "        [1, 2, 3],\n",
+       "        [2, 3, 4],\n",
+       "        [2, 3, 4],\n",
+       "        [1, 2, 3],\n",
+       "        [1, 2, 3],\n",
+       "        [3, 4, 5],\n",
+       "        [1, 2, 3],\n",
+       "        [3, 4, 5],\n",
+       "        [3, 4, 5]]),\n",
+       " 'test_score': array([0.73626374, 0.80619381, 0.77822178, 0.77922078, 0.80619381,\n",
+       "        0.78821179, 0.75224775, 0.77322677, 0.77322677, 0.80819181,\n",
+       "        0.81718282, 0.7982018 , 0.82817183, 0.81618382, 0.85514486,\n",
+       "        0.84615385, 0.84515485]),\n",
+       " 'model_id': array(['bracket=2-0', 'bracket=2-1', 'bracket=2-2', 'bracket=2-3',\n",
+       "        'bracket=2-4', 'bracket=2-5', 'bracket=2-6', 'bracket=2-7',\n",
+       "        'bracket=2-8', 'bracket=1-0', 'bracket=1-1', 'bracket=1-2',\n",
+       "        'bracket=1-3', 'bracket=1-4', 'bracket=0-0', 'bracket=0-1',\n",
+       "        'bracket=0-2'], dtype='<U11'),\n",
+       " 'bracket': array([2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 0, 0, 0]),\n",
+       " 'rank_test_score': array([9, 1, 5, 4, 1, 3, 8, 6, 6, 4, 2, 5, 1, 3, 1, 2, 3]),\n",
+       " 'std_score_time': array([0.        , 0.00367966, 0.        , 0.        , 0.00024688,\n",
+       "        0.01542246, 0.        , 0.        , 0.        , 0.00037813,\n",
+       "        0.00196266, 0.0007596 , 0.00154793, 0.0058749 , 0.00411808,\n",
+       "        0.01123738, 0.00584662]),\n",
+       " 'params': array([{'batch_size': 32, 'module__dropout': 0.14233276095945843, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 100},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.15454961327883227, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 100},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.2735040006927046, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 25},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.2644422586210435, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 50},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.13400896038886034, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 25},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.1930240908469314, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 25},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.1514203365611017, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 25},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.23147478210838274, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 50},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.2851643369137857, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 50},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.15385363146857528, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 25},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.18147304066270142, 'module__filter_sizes': (2, 3, 4), 'module__n_filters': 100},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.29957987553007753, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 25},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.10990512595594133, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 100},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.2900391400227389, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 100},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.12476236679704862, 'module__filter_sizes': (1, 2, 3), 'module__n_filters': 50},\n",
+       "        {'batch_size': 64, 'module__dropout': 0.14516028378189222, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 100},\n",
+       "        {'batch_size': 32, 'module__dropout': 0.10459089167362097, 'module__filter_sizes': (3, 4, 5), 'module__n_filters': 100}],\n",
+       "       dtype=object),\n",
+       " 'mean_partial_fit_time': array([3.08231282, 4.20376865, 1.61370659, 1.77556491, 2.55069685,\n",
+       "        2.75940621, 1.6858294 , 1.94340801, 1.67727423, 2.52923703,\n",
+       "        4.61230862, 2.45016968, 3.72088039, 4.44598476, 3.63528967,\n",
+       "        6.53302658, 6.50973403]),\n",
+       " 'mean_score_time': array([0.16500354, 0.17686621, 0.05370951, 0.07551599, 0.06296408,\n",
+       "        0.08132923, 0.06409264, 0.09241557, 0.07140708, 0.0622654 ,\n",
+       "        0.17359829, 0.05686355, 0.13362586, 0.17294081, 0.06504714,\n",
+       "        0.17865801, 0.17236865]),\n",
+       " 'partial_fit_calls': array([ 1,  9,  1,  1,  3,  3,  1,  1,  1,  4,  4,  4,  4, 12, 12, 12, 12])}"
       ]
      },
-     "execution_count": 82,
+     "execution_count": 64,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "y.dtype"
+    "search.cv_results_"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 81,
+   "execution_count": 65,
    "metadata": {},
    "outputs": [
     {
-     "ename": "KeyboardInterrupt",
-     "evalue": "",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-81-61fb781fe4aa>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# why did this take so long and why is the GPU running?\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0my\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask/base.py\u001b[0m in \u001b[0;36mcompute\u001b[0;34m(self, **kwargs)\u001b[0m\n\u001b[1;32m    164\u001b[0m         \u001b[0mdask\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbase\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompute\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    165\u001b[0m         \"\"\"\n\u001b[0;32m--> 166\u001b[0;31m         \u001b[0;34m(\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraverse\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    167\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    168\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/dask/base.py\u001b[0m in \u001b[0;36mcompute\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    442\u001b[0m         \u001b[0mpostcomputes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__dask_postcompute__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    443\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 444\u001b[0;31m     \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mschedule\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdsk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkeys\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    445\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mrepack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpostcomputes\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    446\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36mget\u001b[0;34m(self, dsk, keys, restrictions, loose_restrictions, resources, sync, asynchronous, direct, retries, priority, fifo_timeout, actors, **kwargs)\u001b[0m\n\u001b[1;32m   2664\u001b[0m                     \u001b[0mshould_rejoin\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2665\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2666\u001b[0;31m                 \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgather\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpacked\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0masynchronous\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0masynchronous\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdirect\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdirect\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2667\u001b[0m             \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2668\u001b[0m                 \u001b[0;32mfor\u001b[0m \u001b[0mf\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mfutures\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36mgather\u001b[0;34m(self, futures, errors, direct, asynchronous)\u001b[0m\n\u001b[1;32m   1965\u001b[0m                 \u001b[0mdirect\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdirect\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1966\u001b[0m                 \u001b[0mlocal_worker\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mlocal_worker\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1967\u001b[0;31m                 \u001b[0masynchronous\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0masynchronous\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1968\u001b[0m             )\n\u001b[1;32m   1969\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/client.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(self, func, asynchronous, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    814\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    815\u001b[0m             return sync(\n\u001b[0;32m--> 816\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback_timeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcallback_timeout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    817\u001b[0m             )\n\u001b[1;32m    818\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/site-packages/distributed/utils.py\u001b[0m in \u001b[0;36msync\u001b[0;34m(loop, func, callback_timeout, *args, **kwargs)\u001b[0m\n\u001b[1;32m    342\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    343\u001b[0m         \u001b[0;32mwhile\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mis_set\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 344\u001b[0;31m             \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    345\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m         \u001b[0mtyp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0merror\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    550\u001b[0m             \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flag\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    551\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 552\u001b[0;31m                 \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_cond\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    553\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    554\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;32m/opt/conda/lib/python3.7/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    298\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    299\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 300\u001b[0;31m                     \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    301\u001b[0m                 \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    302\u001b[0m                     \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
-      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Data must be 1-dimensional\n"
      ]
     }
    ],
    "source": [
-    "# why isn't this responding and why is the GPU running?\n",
-    "y[:2].compute()"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## Integration"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "`HyperbandSearchCV` follows the Scikit-learn API and mirrors Scikit-learn's `RandomizedSearchCV`. This means that it \"just works\". All the Scikit-learn attributes and methods are available:"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "search.best_score_"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "search.best_estimator_"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import pandas as pd"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "search.cv_results_"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "cv_results = pd.DataFrame(search.cv_results_)\n",
-    "cv_results.head()"
+    "# issue with numpy converting array of tuples into a 2d array\n",
+    "try:    \n",
+    "    cv_results = pd.DataFrame(search.cv_results_)\n",
+    "    cv_results.head()\n",
+    "except Exception as e:\n",
+    "    print(e)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 66,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1673,9 +1317,178 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 67,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>param_module__n_filters</th>\n",
+       "      <th>param_batch_size</th>\n",
+       "      <th>param_module__dropout</th>\n",
+       "      <th>std_partial_fit_time</th>\n",
+       "      <th>param_module__filter_sizes</th>\n",
+       "      <th>test_score</th>\n",
+       "      <th>model_id</th>\n",
+       "      <th>bracket</th>\n",
+       "      <th>rank_test_score</th>\n",
+       "      <th>std_score_time</th>\n",
+       "      <th>params</th>\n",
+       "      <th>mean_partial_fit_time</th>\n",
+       "      <th>mean_score_time</th>\n",
+       "      <th>partial_fit_calls</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>100</td>\n",
+       "      <td>32</td>\n",
+       "      <td>0.142333</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>[3, 4, 5]</td>\n",
+       "      <td>0.736264</td>\n",
+       "      <td>bracket=2-0</td>\n",
+       "      <td>2</td>\n",
+       "      <td>9</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.142332...</td>\n",
+       "      <td>3.082313</td>\n",
+       "      <td>0.165004</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>100</td>\n",
+       "      <td>64</td>\n",
+       "      <td>0.154550</td>\n",
+       "      <td>1.427928</td>\n",
+       "      <td>[2, 3, 4]</td>\n",
+       "      <td>0.806194</td>\n",
+       "      <td>bracket=2-1</td>\n",
+       "      <td>2</td>\n",
+       "      <td>1</td>\n",
+       "      <td>0.003680</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.154549...</td>\n",
+       "      <td>4.203769</td>\n",
+       "      <td>0.176866</td>\n",
+       "      <td>9</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>25</td>\n",
+       "      <td>64</td>\n",
+       "      <td>0.273504</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>[1, 2, 3]</td>\n",
+       "      <td>0.778222</td>\n",
+       "      <td>bracket=2-2</td>\n",
+       "      <td>2</td>\n",
+       "      <td>5</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.273504...</td>\n",
+       "      <td>1.613707</td>\n",
+       "      <td>0.053710</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>50</td>\n",
+       "      <td>32</td>\n",
+       "      <td>0.264442</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>[2, 3, 4]</td>\n",
+       "      <td>0.779221</td>\n",
+       "      <td>bracket=2-3</td>\n",
+       "      <td>2</td>\n",
+       "      <td>4</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.264442...</td>\n",
+       "      <td>1.775565</td>\n",
+       "      <td>0.075516</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>25</td>\n",
+       "      <td>32</td>\n",
+       "      <td>0.134009</td>\n",
+       "      <td>0.888979</td>\n",
+       "      <td>[2, 3, 4]</td>\n",
+       "      <td>0.806194</td>\n",
+       "      <td>bracket=2-4</td>\n",
+       "      <td>2</td>\n",
+       "      <td>1</td>\n",
+       "      <td>0.000247</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.134008...</td>\n",
+       "      <td>2.550697</td>\n",
+       "      <td>0.062964</td>\n",
+       "      <td>3</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   param_module__n_filters  param_batch_size  param_module__dropout  \\\n",
+       "0                      100                32               0.142333   \n",
+       "1                      100                64               0.154550   \n",
+       "2                       25                64               0.273504   \n",
+       "3                       50                32               0.264442   \n",
+       "4                       25                32               0.134009   \n",
+       "\n",
+       "   std_partial_fit_time param_module__filter_sizes  test_score     model_id  \\\n",
+       "0              0.000000                  [3, 4, 5]    0.736264  bracket=2-0   \n",
+       "1              1.427928                  [2, 3, 4]    0.806194  bracket=2-1   \n",
+       "2              0.000000                  [1, 2, 3]    0.778222  bracket=2-2   \n",
+       "3              0.000000                  [2, 3, 4]    0.779221  bracket=2-3   \n",
+       "4              0.888979                  [2, 3, 4]    0.806194  bracket=2-4   \n",
+       "\n",
+       "   bracket  rank_test_score  std_score_time  \\\n",
+       "0        2                9        0.000000   \n",
+       "1        2                1        0.003680   \n",
+       "2        2                5        0.000000   \n",
+       "3        2                4        0.000000   \n",
+       "4        2                1        0.000247   \n",
+       "\n",
+       "                                              params  mean_partial_fit_time  \\\n",
+       "0  {'batch_size': 32, 'module__dropout': 0.142332...               3.082313   \n",
+       "1  {'batch_size': 64, 'module__dropout': 0.154549...               4.203769   \n",
+       "2  {'batch_size': 64, 'module__dropout': 0.273504...               1.613707   \n",
+       "3  {'batch_size': 32, 'module__dropout': 0.264442...               1.775565   \n",
+       "4  {'batch_size': 32, 'module__dropout': 0.134008...               2.550697   \n",
+       "\n",
+       "   mean_score_time  partial_fit_calls  \n",
+       "0         0.165004                  1  \n",
+       "1         0.176866                  9  \n",
+       "2         0.053710                  1  \n",
+       "3         0.075516                  1  \n",
+       "4         0.062964                  3  "
+      ]
+     },
+     "execution_count": 67,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "cv_results = pd.DataFrame(search.cv_results_)\n",
     "cv_results.head()"
@@ -1683,27 +1496,98 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 68,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "0.8106"
+      ]
+     },
+     "execution_count": 68,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.score(X_test, y_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 69,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<table>\n",
+       "<tr>\n",
+       "<td>\n",
+       "<table>\n",
+       "  <thead>\n",
+       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr><th> Bytes </th><td> 40.00 kB </td> <td> 40.00 kB </td></tr>\n",
+       "    <tr><th> Shape </th><td> (5000,) </td> <td> (5000,) </td></tr>\n",
+       "    <tr><th> Count </th><td> 2 Tasks </td><td> 1 Chunks </td></tr>\n",
+       "    <tr><th> Type </th><td> int64 </td><td> numpy.ndarray </td></tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</td>\n",
+       "<td>\n",
+       "<svg width=\"170\" height=\"75\" style=\"stroke:rgb(0,0,0);stroke-width:1\" >\n",
+       "\n",
+       "  <!-- Horizontal lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"120\" y2=\"0\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"0\" y1=\"25\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Vertical lines -->\n",
+       "  <line x1=\"0\" y1=\"0\" x2=\"0\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "  <line x1=\"120\" y1=\"0\" x2=\"120\" y2=\"25\" style=\"stroke-width:2\" />\n",
+       "\n",
+       "  <!-- Colored Rectangle -->\n",
+       "  <polygon points=\"0.000000,0.000000 120.000000,0.000000 120.000000,25.412617 0.000000,25.412617\" style=\"fill:#ECB172A0;stroke-width:0\"/>\n",
+       "\n",
+       "  <!-- Text -->\n",
+       "  <text x=\"60.000000\" y=\"45.412617\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" >5000</text>\n",
+       "  <text x=\"140.000000\" y=\"12.706308\" font-size=\"1.0rem\" font-weight=\"100\" text-anchor=\"middle\" transform=\"rotate(0,140.000000,12.706308)\">1</text>\n",
+       "</svg>\n",
+       "</td>\n",
+       "</tr>\n",
+       "</table>"
+      ],
+      "text/plain": [
+       "dask.array<_predict, shape=(5000,), dtype=int64, chunksize=(5000,), chunktype=numpy.ndarray>"
+      ]
+     },
+     "execution_count": 69,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.predict(X_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 70,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([0, 1, 0, ..., 1, 0, 1])"
+      ]
+     },
+     "execution_count": 70,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "search.predict(X_test).compute()"
    ]
@@ -1717,9 +1601,128 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 71,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>model_id</th>\n",
+       "      <th>params</th>\n",
+       "      <th>partial_fit_calls</th>\n",
+       "      <th>partial_fit_time</th>\n",
+       "      <th>score</th>\n",
+       "      <th>score_time</th>\n",
+       "      <th>elapsed_wall_time</th>\n",
+       "      <th>bracket</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>bracket=0-0</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.124762...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>3.679612</td>\n",
+       "      <td>0.798202</td>\n",
+       "      <td>0.069165</td>\n",
+       "      <td>30.430613</td>\n",
+       "      <td>0</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>bracket=0-1</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.145160...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>6.269561</td>\n",
+       "      <td>0.791209</td>\n",
+       "      <td>0.167421</td>\n",
+       "      <td>30.430616</td>\n",
+       "      <td>0</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>bracket=0-2</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.104590...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>6.193064</td>\n",
+       "      <td>0.789211</td>\n",
+       "      <td>0.166522</td>\n",
+       "      <td>30.430617</td>\n",
+       "      <td>0</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>bracket=1-0</td>\n",
+       "      <td>{'batch_size': 32, 'module__dropout': 0.153853...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>1.683563</td>\n",
+       "      <td>0.787213</td>\n",
+       "      <td>0.062644</td>\n",
+       "      <td>36.104408</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>bracket=1-1</td>\n",
+       "      <td>{'batch_size': 64, 'module__dropout': 0.181473...</td>\n",
+       "      <td>1</td>\n",
+       "      <td>3.002587</td>\n",
+       "      <td>0.803197</td>\n",
+       "      <td>0.171636</td>\n",
+       "      <td>36.104410</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "      model_id                                             params  \\\n",
+       "0  bracket=0-0  {'batch_size': 32, 'module__dropout': 0.124762...   \n",
+       "1  bracket=0-1  {'batch_size': 64, 'module__dropout': 0.145160...   \n",
+       "2  bracket=0-2  {'batch_size': 32, 'module__dropout': 0.104590...   \n",
+       "3  bracket=1-0  {'batch_size': 32, 'module__dropout': 0.153853...   \n",
+       "4  bracket=1-1  {'batch_size': 64, 'module__dropout': 0.181473...   \n",
+       "\n",
+       "   partial_fit_calls  partial_fit_time     score  score_time  \\\n",
+       "0                  1          3.679612  0.798202    0.069165   \n",
+       "1                  1          6.269561  0.791209    0.167421   \n",
+       "2                  1          6.193064  0.789211    0.166522   \n",
+       "3                  1          1.683563  0.787213    0.062644   \n",
+       "4                  1          3.002587  0.803197    0.171636   \n",
+       "\n",
+       "   elapsed_wall_time  bracket  \n",
+       "0          30.430613        0  \n",
+       "1          30.430616        0  \n",
+       "2          30.430617        0  \n",
+       "3          36.104408        1  \n",
+       "4          36.104410        1  "
+      ]
+     },
+     "execution_count": 71,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "hist = pd.DataFrame(search.history_)\n",
     "hist.head()"
@@ -1750,16 +1753,14 @@
     "\n",
     "Performance comparisons can be found in the SciPy 2019 talk/paper."
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {
+  "environment": {
+   "name": "pytorch-gpu.1-4.m46",
+   "type": "gcloud",
+   "uri": "gcr.io/deeplearning-platform-release/pytorch-gpu.1-4:m46"
+  },
   "kernelspec": {
    "display_name": "Python 3",
    "language": "python",
@@ -1775,7 +1776,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.7.6"
   }
  },
  "nbformat": 4,