lukec

Results 21 comments of lukec

Do you have the results of the accuracy test?

* fix util error when egale config=314 @quinnrong94 > For Future PRs: > > * Do some profiling and check whether there is any bubble caused by synchronization between CPU...

> Do we need raise error for bf16 when enable deepep? I'm not sure, it's necessary? @zhyncs @ch-wan

> expect this feature On the way on the way

> Greate job! If I want to participate in VLM, what can I do? You can contact me on Slack. We have an eagle-vlm team. My slack is Chao Wang....

> Interested in supporting DS V3/R1, who should I reach out to? You can search for specforge in the sgl project of slack.

How much performance improvement does flex attention offer in comparison?

Could you fix the code format @ValeGian

Great job!!!!! This enables us to support all MLLM models. @FrankLeeeee

Is the training speed improved compared to the original implementation?