🧠 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
Merge pull request #912 from Afloat16/fix/generation-greedy-temperature
eaa4b87 jingyaogong committed 1 day ago
Merge pull request #909 from Afloat16/fix/distillation-empty-supervision
5d8981e jingyaogong committed 1 day ago
Merge pull request #907 from Afloat16/fix/attention-bool-mask
2bbf372 jingyaogong committed 1 day ago
Merge pull request #905 from Afloat16/fix/api-inference-event-loop
2794acc jingyaogong committed 1 day ago
Merge pull request #904 from FeiyangP/fix/agent-reward-numeric-gt
12b5e03 jingyaogong committed 1 day ago
[fix] LoRA: add alpha/rank scaling with backward-compatible persistence
9012e5e uphomer committed 16 days ago
Merge pull request #883 from cyanseek/codex/stream-tool-content
d80c692 jingyaogong committed 23 days ago
Merge pull request #886 from Yi-111-a/docs/grpo-policy-loss-interpretation
7a47668 jingyaogong committed 23 days ago
Merge pull request #894 from BBqBuleberry/feat/lora-rank-arg
0938214 jingyaogong committed 23 days ago
Merge pull request #882 from Yi-111-a/fix/train-agent-malformed-tool-call
6bc64cd jingyaogong committed 23 days ago
Merge pull request #889 from fjvbn2003/fix-load-from-model-dir
999dfde jingyaogong committed 23 days ago
Merge pull request #896 from Vincent-Huang-2000/fix/resume-ckpt-fp32
7ea219f jingyaogong committed 23 days ago
feat(lora): make LoRA rank configurable in train_lora.py and eval_llm.py
695aa1f BBqBuleberry committed 25 days ago
[fix] decide native-vs-transformers loading by directory name, not substring
cd36b4a YoungjuKim committed 1 month ago
[fix] keep malformed tool calls from crashing agent RL training
dcbd31b Yi-111-a committed 1 month ago