203 Star 1.3K Fork 1.2K

GVPAscend/MindSpeed-LLM
暂停

增加调用mindspeed中负载均衡特性功能 可合并
ci-pipeline-failed
ascend-cla/yes
!3201 zhanggaolu 11
审查:
2025-08-26 09:41
!3088 aaron 1
审查:
2025-08-04 11:31
【轻量级 PR】:add docs/pytorch/features. 可合并
ci-pipeline-failed
ascend-cla/yes
!3068 aaron 2
审查:
2025-07-30 14:58
add examples/mcore/qwen3_moe/pretrain_qwen3_30b_a3b_4K_ptd_mamba.sh. 存在冲突
ascend-cla/no
stat/needs-squash
!3038 flippy航 2
审查:
2025-07-21 17:51
feat: balanced moe 存在冲突
ci-pipeline-failed
ascend-cla/yes
!2856 邓佳 11
审查:
2025-06-19 14:38
optimize when CP and variable seq length functions are both enabled 可合并
ci-pipeline-passed
ascend-cla/yes
stat/needs-squash
!2853 xiecheng 40
审查:
2025-06-19 10:55
The get_batch performance is optimized. 存在冲突
ci-pipeline-failed
ascend-cla/yes
stat/needs-squash
!2641 aaron 26
审查:
2025-05-07 00:15
!2574 吴量文 1
审查:
2025-04-22 14:24
客户支撑,RL训练评测集测试 存在冲突
ascend-cla/yes
!2446 朱意霖 6
审查:
2025-03-23 11:07
Python
1
https://gitee.com/ascend/MindSpeed-LLM.git
git@gitee.com:ascend/MindSpeed-LLM.git
ascend
MindSpeed-LLM
MindSpeed-LLM

搜索帮助