tmac.liang

user： 最后一行的用户被忽略； crontab：不规范的空格导致panic listen：进程名为空时导致panic

**Describe the bug** we find cudaMemcpyAsync run too much time on torch profiler: 1. on aten::_local_scalar_dense  2. on _has_inf_or_nan  **Expected behavior** reduce cudaMemcpyAsync op **ds_report output** ``` [2023-12-25...

HI，Is there plan to upgrade megatron-lm，it can be support more feature and better performance

Are there any plans to support context parallel?

5

Long-text scenarios are quite common, and it would be of great help if they could be supported.