use pdm instead, new build process

Signed-off-by: Alex Chi Z <[email protected]>

Author: skyzh

.clang-format

@@ -0,0 +1,6 @@
+BasedOnStyle: Google
+DerivePointerAlignment: false
+PointerAlignment: Right
+ColumnLimit: 120
+IndentWidth: 4
+AccessModifierOffset: -4

.gitignore

@@ -172,3 +172,6 @@ cython_debug/
 
 # PyPI configuration file
 .pypirc
+
+*.dylib
+*.metallib

.vscode/settings.json

@@ -0,0 +1,6 @@
+{
+    "clangd.arguments": [
+        "--compile-commands-dir=${workspaceFolder}/src/extensions_ref/build/temp.macosx-15.0-arm64-cpython-312/tiny_llm_ext_ref._ext"
+    ],
+    "cmake.ignoreCMakeListsMissing": true
+}

book/src/setup.md

@@ -1,12 +1,10 @@
 # Setting Up the Environment
 
-To follow along this course, you will need a mactonish device with Apple Silicon. We manage the codebase with the poetry
-dependency manager.
+To follow along this course, you will need a mactonish device with Apple Silicon. We manage the codebase with pdm.
 
-## Install Poetry
+## Install pdm
 
-Please follow the [offcial guide](https://python-poetry.org/docs/#installing-with-the-official-installer) to install
-poetry.
+Please follow the [offcial guide](https://pdm-project.org/en/latest/) to install pdm.
 
 ## Clone the Repository
 
@@ -30,23 +28,23 @@ We provide all reference implementations and you can refer to them if you get st
 
 ```bash
 cd tiny-llm
-poetry install
+pdm install -v # this will automatically create a virtual environment and install all dependencies
 ```
 
 ## Check the Installation
 
 ```bash
-poetry run python check.py
+pdm run python check.py
 # The reference solution should pass all the tests
-poetry run pytest tests_ref_impl_week1
+pdm run pytest tests_ref_impl_week1
 ```
 
 ## Run Unit Tests
 
 Your code is in `src/tiny_llm`. You can run the unit tests with:
 
 ```bash
-poetry run pytest tests
+pdm run pytest tests
 ```
 
 ## Download the Model Parameters
@@ -58,21 +56,21 @@ after week 1 day 6 when you start to load the model parameters.)
 
 Follow the guide of [this page](https://huggingface.co/docs/huggingface_hub/main/en/guides/cli) to install the huggingface
 cli. You should install it in your user directory/globally instead of in the tiny-llm virtual environment created by
-poetry.
+pdm.
 
 The model parameters are hosted on Hugging Face. Once you authenticated your cli with the credentials, you can download
 them with:
 
 ```bash
-# do not do this in the virtual environment created by poetry; do `deactivate` first if you did `poetry shell`
+# do not do this in the virtual environment created by pdm
 huggingface-cli login
 huggingface-cli download Qwen/Qwen2-7B-Instruct-MLX
 ```
 
 Then, you can run:
 
 ```bash
-poetry run python main_ref_impl_week1.py
+pdm run python main_ref_impl_week1.py
 ```
 
 It should load the model and print some text.

book/src/week1-01-attention.md

@@ -25,7 +25,7 @@ we will pass a tensor of the shape `N.. x 1024 x 512` to the attention layer.
 In this task, we will implement the scaled dot product attention function.
 
 ```
-poetry run pytest tests -k week_1_day_1_task_1 -v
+pdm run pytest tests -k week_1_day_1_task_1 -v
 ```
 
 
@@ -66,8 +66,8 @@ mask: 1 x H x L x L
 At the end of this task, you should be able to pass the following tests:
 
 ```
-poetry run pytest tests -k test_attention_simple
-poetry run pytest tests -k test_attention_with_mask
+pdm run pytest tests -k test_attention_simple
+pdm run pytest tests -k test_attention_with_mask
 ```
 
 ## Task 2: Implement `MultiHeadAttention`
@@ -115,7 +115,7 @@ W_o: (H x D) x E
 At the end of the day, you should be able to pass the following tests:
 
 ```
-poetry run pytest tests -k week_1_day_1_task_2 -v
+pdm run pytest tests -k week_1_day_1_task_2 -v
 ```
 
 {{#include copyright.md}}

book/src/week1-02-positional-encodings.md

@@ -52,7 +52,7 @@ You can do this by reshaping `x` to (N, L, H, D // 2, 2) and then applying the a
 You can test your implementation by running the following command:
 
 ```
-poetry run pytest tests -k week_1_day_2_task_1 -v
+pdm run pytest tests -k week_1_day_2_task_1 -v
 ```
 
 ## Task 2: Implement `RoPE` in the non-traditional form
@@ -74,7 +74,7 @@ frequencies to each half separately.
 You can test your implementation by running the following command:
 
 ```
-poetry run pytest tests -k week_1_day_2_task_2 -v
+pdm run pytest tests -k week_1_day_2_task_2 -v
 ```
 
 **📚 Readings**

book/src/week2-overview.md

@@ -2,3 +2,5 @@ https://github.com/ml-explore/mlx/blob/main/mlx/backend/cpu/quantized.cpp
 https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/layers/linear.py
 MLX uses INT4 W4A16
 https://ml-explore.github.io/mlx/build/html/dev/extensions.html
+
+pdm run ./build_ext.sh

build-extension.py

@@ -0,0 +1,4 @@
+#!/bin/bash
+
+cd src/extensions_ref
+python setup.py build_ext --inplace