/nanogpt

Implementation of GPT in pytorch 🤖

Primary LanguagePythonMIT LicenseMIT

nanogpt

Implementation of nanogpt from karpathy. I've modified this model to text classification by chopping off the language model head and adding a binary classifier on top.

Development

poetry install

Training

Test run:

python train.py --config-name rotten_tomatoes_binary_classification_fast

Full training run:

python train.py --config-name rotten_tomatoes_binary_classification