/
rnekrasov
/
google-research
GigaCode
beta
ОбзорЦентр заботыВойти

google-research

Форк
0

Ветки: 311Коммиты: 4490Теги: 0
..
google-research
/
sudoku_gpt 
inference
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
README.md
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
data.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
evaluater.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
main.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
model.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
othello.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
requirements.txt
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
sudoku.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
train_and_evaluate.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
trainer.py
Experiments related to the project on learning to play Sudoku using transformer models trained using next-token prediction.
6 месяцев назад
README.md

Library for training a GPT model on Sudoku puzzles.

This document describes the functionality and experiments in this project. We train decoder-only Transformer language models on Sudoku puzzles and their solutions and study how well the model learns to solve unseen Sudoku puzzles at test time. In addition, we perform an interpretability analysis on the internal embeddings of the model to study whether it has developed an in-depth understanding of the rules and strategies for solving Sudoku.

From google-research/

To run the experiment, run

python -m sudoku_gpt/main.py

License

Sudoku-GPT is licensed under the Apache License, Version 2.0. This is not an officially supported Google product. Contact nishanthd@google.com for questions about the repo.

Использование cookies

Мы используем файлы cookie в соответствии с Политикой конфиденциальности и Политикой использования cookies.

Нажимая кнопку «Принимаю», Вы даете АО «СберТех» согласие на обработку Ваших персональных данных в целях совершенствования нашего веб-сайта и Сервиса GitVerse, а также повышения удобства их использования.

Запретить использование cookies Вы можете самостоятельно в настройках Вашего браузера.