Skip to content

🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

Commits on master

0f6803e
[update] remove expert args
jingyaogong committed 2 months ago
6fc918b
4bd6062
[update] 把随机种子暴露到命令行
Justin Guese committed 2 months ago
fc585a9
[update] readme
jingyaogong committed 2 months ago
7a6fddd
[update] dependency version
jingyaogong committed 2 months ago
c9a8ab4
56d4cdd
[update] dependency version
jingyaogong committed 2 months ago
e841568
[update] readme content
jingyaogong committed 3 months ago
d65ef2c
[update] readme content
jingyaogong committed 3 months ago
ed5ea9e
[update] readme content
jingyaogong committed 3 months ago
eabad70
[update] supporter charts
jingyaogong committed 3 months ago
393e387
[refactor] PPO cleanup
jingyaogong committed 4 months ago
89d674b
[fix] PPO rollout mode
jingyaogong committed 4 months ago
2d4c385
[fix] robustness
jingyaogong committed 4 months ago
512eed0
[fix] ddp exit hang
jingyaogong committed 5 months ago
3f1a7cc
[fix] lora dims
jingyaogong committed 5 months ago
4a68da7
[fix] issue#771
jingyaogong committed 6 months ago
4497610
182bff1
[update] requirements
jingyaogong committed 6 months ago
9da8e1a
[fix] repetition_penalty
jingyaogong committed 6 months ago
dddedc6
[feat] reduce RL memory
jingyaogong committed 6 months ago
802c15b
1077641