simple-stories-train 0.0.1

Training of small LM models for SimpleStories

simple_stories_train

Project for training small LMs. Designed for training on SimpleStories, an extension of TinyStories.

• Training script is based on the efficeint train_gpt2.py in llm.c (licensed under MIT ((c) 2024 Andrei Karpathy))
• Some model architecture implementations are based on TransformerLens (licensed under MIT ((c) 2022 TransformerLensOrg)).

Installation

From the root of the repository, run one of

make install-dev  # To install the package, dev requirements and pre-commit hooks
make install  # To just install the package (runs `pip install -e .`)

Development

Suggested extensions and settings for VSCode are provided in .vscode/. To use the suggested settings, copy .vscode/settings-example.json to .vscode/settings.json.

There are various make commands that may be helpful

make check  # Run pre-commit on all files (i.e. pyright, ruff linter, and ruff formatter)
make type  # Run pyright on all files
make format  # Run ruff linter and formatter on all files
make test  # Run tests that aren't marked `slow`
make test-all  # Run all tests

Usage

Training a model

python train_llama.py [PATH/TO/CONFIG.yaml] [--key1 value1 --key2 value2 ...]

where

• PATH/TO/CONFIG.yaml contains the training config. If no path is provided, a default config will be used.
• --key1 value1 --key2 value2 ... override values in the config. Note that if you wish to update a nested value, you must use dotted notation (e.g. --train_dataset_config.name my_dataset).

If running on CPU, you may need to set --compile=False.

To run on multiple GPUs, use

torchrun --standalone --nproc_per_node=N train_llama.py ...

where N is the number of GPUs to use.

Logging with Weights & Biases

To track training with Weights & Biases, you can set the WANDB_PROJECT and WANDB_API_KEY variables in .env. API keys can be obtained from your Weights & Biases account settings.