🚀🚀 「大模型」2小时完全从0训练64M的小参数GPT!🌏 Train a 64M-parameter GPT from scratch in just 2h!
COMMITS
June 1, 2026
J
[fix] ddp exit hang
jingyaogong committed
May 31, 2026
J
[fix] lora dims
jingyaogong committed
May 19, 2026
J
[fix] issue#771
jingyaogong committed
May 16, 2026
J
[update] requirements
jingyaogong committed
May 7, 2026
J
[fix] repetition_penalty
jingyaogong committed
May 6, 2026
J
[feat] reduce RL memory
jingyaogong committed
J
Merge pull request #759 from TKiteRunner/fix/grpo-oom-reorder-rewards
jingyaogong committed
T
fix: resolve CUDA OOM in train_grpo.py on GPUs with <=8GB VRAM
TKiteRunner committed
J
[update] readme
jingyaogong committed
May 3, 2026
J
[fix] inference bug
jingyaogong committed
May 1, 2026
J
[update] readme
jingyaogong committed
April 28, 2026
J
[update] readme
jingyaogong committed
April 27, 2026
J
[fix] bugs
jingyaogong committed
J
[fix] rollout bugs
jingyaogong committed
J
[update] readme
jingyaogong committed
April 24, 2026
J
[update] readme
jingyaogong committed
J
[update] readme
jingyaogong committed
April 21, 2026
J
[update] readme
jingyaogong committed
J
[fix] bugs
jingyaogong committed
April 19, 2026
J
[fix] transformers-5.x
jingyaogong committed
J
[update] tie embedding
jingyaogong committed
J
[update] readme
jingyaogong committed
April 10, 2026
J
[update] readme
jingyaogong committed
April 9, 2026
J
[fix] lora compile
jingyaogong committed
J
[update] readme
jingyaogong committed
J
[update] readme
jingyaogong committed
J
[update] bench
jingyaogong committed
J
[fix] lora moe
jingyaogong committed
J
[update] add a comment
jingyaogong committed
April 4, 2026
J
[update] image
jingyaogong committed