2.2.0

ascend-robot[pytorch][bugfix]fix deepseek3 tnd bug in mbs > 1

031c12d9创建于 2月27日历史提交

文件	最后提交记录	最后更新时间
checkpoint	[pytorch][bugfix]fix -mla-mm-split in ckpt	4 个月前
common	!2569 Refact: embedding and rotary embedding	1 年前
dataset	!2710 Compatibility fixes	1 年前
evaluation	[pytorch][bugfix]fix about cleancode	8 个月前
inference	!2712 [pytorch][feature]upgrading Megatron to r0.12.1	11 个月前
megatron_basic	[pytorch][bugfix]fix lora_target_modules in ckpt save_lora_to_hf	4 个月前
models	[pytorch][bugfix]fix deepseek3 tnd bug in mbs > 1	3 个月前
posttrain	[pytorch][bugfix] fix variable in lora moe	4 个月前
preprocess	!3355 [pytroch][bugfix] fix bug when using GeneralPretrainHandler to process shareGPT-style datasets	8 个月前
utils	!1958 整改仓库文件结构	1 年前
__init__.py	!1998 rename: repo package name from modellink to mindspeed_llm	1 年前
megatron_adaptor.py	!3100 [pytorch][feature]mindio_tft adapt reusefp32 and megatron_adaptor_v2	9 个月前
megatron_adaptor_v2.py	!3100 [pytorch][feature]mindio_tft adapt reusefp32 and megatron_adaptor_v2	9 个月前