Star182
226
代码介绍
代码
Issues71
Pull Requests56
流水线
Actions
讨论
Wiki
项目成员54
分析
项目设置
Star182
226
  1. MindSpeed-LLM
  2. /
  3. mindspeed_llm
  4. /
  5. features_manager
  6. /
  7. megatron_basic
ascend-robotascend-robotfix(pytorch):Ensure no PP/VPP stage contains only empty layers during LoRA fine-tuning
337813b0创建于 4月2日历史提交
文件最后提交记录最后更新时间
__init__.py
!3075 [pytorch][refactor]refactor pipeline and te features10 个月前
megatron_basic.py
[pytorch][bugfix] fix reapeated allgather in 2.2.07 个月前
model_basic.py
!3114 [pytorch][refactor] refactor context parallel9 个月前
requirements_basic.py
[pytorch][bugfix] update checkpoint loading for LoRA and QLoRA7 个月前
training_basic.py
fix(pytorch):Ensure no PP/VPP stage contains only empty layers during LoRA fine-tuning2 个月前