yue

icon indicating a website link https://kangqiyue.github.io/

icon company Westlake University icon location Westlake University, Posdoc

Results 5 comments of yue

apply transformation using `VideoReader`

I write a transform to do the transformation. I think we can read the audio in its original shape, then use torchvision to do this. `from torchvision import transforms as...

Fix mid block checker to fix custom model loading

Thanks! cool !

Question about transformer support

same error, any suggestion?

Question about transformer support

> same error, any suggestion? the version of transformers may be very important, 4.46.3 is OK. other version may lead to nan loss (in my exp). > pip install transformers==4.46.3

[Bug] Unable to fix model output

> @kangqiyue Could you provide a reproducible script? SGLang performs almost identically to `transformers` on MMLU, GSM8K, and Human Eval. We did not observe the performance issues you mentioned. @zhyncs...