bert-base-finetune-3epochs.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "e6966cec-eee9-462b-834a-61f5d43b662d",
   "metadata": {},
   "source": [
    "# Bert Finetune for hatespeech detection task\n",
    "---\n",
    "Dataset: HateSpeechDatasetBalanced.csv   ______________________     Model: Bert Base uncased _________________          epochs: 3   \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "d0753f8c-2840-4ba3-ae3b-9842bff3c38f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Map: 100%|████████████████████████████████████████████████| 726119/726119 [00:10<00:00, 69278.09 examples/s]\n",
      "Map: 100%|█████████████████████████████████████████████████| 726119/726119 [03:35<00:00, 3370.51 examples/s]\n"
     ]
    }
   ],
   "source": [
    "# ai\n",
    "import torch\n",
    "import pandas as pd\n",
    "from datasets import Dataset\n",
    "from transformers import BertTokenizer, BertForSequenceClassification, TrainingArguments, Trainer\n",
    "\n",
    "# Load dataset\n",
    "df = pd.read_csv(\"HateSpeechDatasetBalanced.csv\")  \n",
    "\n",
    "# Load tokenizer\n",
    "tokenizer = BertTokenizer.from_pretrained(\"bert-base-uncased\")\n",
    "\n",
    "# Tokenization function\n",
    "def tokenize_function(examples):\n",
    "    return tokenizer(examples[\"Content\"], padding=\"max_length\", truncation=True, max_length=128)\n",
    "\n",
    "# Convert dataset to Hugging Face format\n",
    "dataset = Dataset.from_pandas(df)\n",
    "\n",
    "# Rename \"Label\" to \"labels\"\n",
    "dataset = dataset.map(lambda x: {\"labels\": x[\"Label\"]}, remove_columns=[\"Label\"])\n",
    "\n",
    "# Tokenize dataset\n",
    "tokenized_dataset = dataset.map(tokenize_function, batched=True)\n",
    "\n",
    "# Train-test split\n",
    "tokenized_dataset = tokenized_dataset.train_test_split(test_size=0.1)\n",
    "train_dataset = tokenized_dataset[\"train\"]\n",
    "val_dataset = tokenized_dataset[\"test\"]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e9b91ec7-a891-4bcd-9a7f-bd38859121ee",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at bert-base-uncased and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "/home/atharva/anaconda3/envs/venv/lib/python3.9/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "      \n",
       "      <progress value='30633' max='30633' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      [30633/30633 1:01:10, Epoch 2/3]\n",
       "    </div>\n",
       "    <table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       " <tr style=\"text-align: left;\">\n",
       "      <th>Epoch</th>\n",
       "      <th>Training Loss</th>\n",
       "      <th>Validation Loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.220600</td>\n",
       "      <td>0.216841</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.066700</td>\n",
       "      <td>0.245783</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploading ./bert-hate-speech.zip to Google Drive folder: Bert finetuned/\n",
      "Upload failed: Command '['rclone', 'copy', './bert-hate-speech.zip', 'gdrive:/Bert finetuned//']' returned non-zero exit status 1.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2025/03/13 09:38:39 Failed to create file system for \"gdrive:/Bert finetuned//\": didn't find section in config file\n"
     ]
    }
   ],
   "source": [
    "# ai\n",
    "import torch\n",
    "import pandas as pd\n",
    "from datasets import Dataset\n",
    "from transformers import BertTokenizer, BertForSequenceClassification, TrainingArguments, Trainer\n",
    "import shutil\n",
    "import subprocess\n",
    "\n",
    "\n",
    "# Load pre-trained BERT model\n",
    "model = BertForSequenceClassification.from_pretrained(\"bert-base-uncased\", num_labels=2)\n",
    "model.gradient_checkpointing_enable()\n",
    "\n",
    "\n",
    "output_dir = \"./bert-hate-speech2\"\n",
    "\n",
    "training_args = TrainingArguments(\n",
    "    output_dir=output_dir,\n",
    "    evaluation_strategy=\"epoch\",\n",
    "    save_strategy=\"epoch\",\n",
    "    per_device_train_batch_size=32,\n",
    "    per_device_eval_batch_size=32,\n",
    "    num_train_epochs=3,\n",
    "    weight_decay=0.01,\n",
    "    bf16=True,\n",
    "    gradient_accumulation_steps=2,\n",
    "    optim=\"adamw_torch_fused\",\n",
    "    logging_dir=\"./logs\",\n",
    "    logging_steps=200,\n",
    "    report_to=\"none\",\n",
    "    save_total_limit=2,\n",
    "    warmup_ratio=0.1,\n",
    "    lr_scheduler_type=\"cosine\",\n",
    "    fp16_full_eval=True,\n",
    "    remove_unused_columns=False,\n",
    ")\n",
    "\n",
    "# Trainer\n",
    "trainer = Trainer(\n",
    "    model=model,\n",
    "    args=training_args,\n",
    "    train_dataset=train_dataset,\n",
    "    eval_dataset=val_dataset,\n",
    ")\n",
    "\n",
    "# Start training\n",
    "trainer.train()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "897c1a79-de82-4f11-b079-b5a621adc879",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "\n",
    "# Extract log history from Trainer\n",
    "log_history = trainer.state.log_history\n",
    "\n",
    "# Convert to DataFrame\n",
    "log_df = pd.DataFrame(log_history)\n",
    "\n",
    "# Keep only loss and accuracy values\n",
    "log_df = log_df.dropna(subset=[\"epoch\"], how=\"any\")\n",
    "\n",
    "# Plot Training Loss & Validation Loss\n",
    "plt.figure(figsize=(20, 5))\n",
    "\n",
    "# Loss Plot\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(log_df[\"epoch\"], log_df[\"loss\"], label=\"Train Loss\", marker=\"o\", linestyle=\"-\", color=\"blue\")\n",
    "plt.plot(log_df[\"epoch\"], log_df[\"eval_loss\"], label=\"Validation Loss\", marker=\"s\", linestyle=\"-\", color=\"red\")\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"Loss Over Epochs\")\n",
    "plt.legend()\n",
    "plt.grid(True, linestyle=\"--\", alpha=0.6)\n",
    "\n",
    "\n",
    "\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0df96e1-6db5-4bdd-87b9-4378924c0b3d",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.21"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}