seanxcwang

tencent shanghai

Results 14 comments of


                                            seanxcwang

[Bug]: with `worker_use_ray = true`, and tensor_parallel_size > 1, the process is pending forever

got the same issue，I solved it by updating accelerate from 0.26.0 to 0.30.0

Add Encoder-decoder model support and T5 Model support

The issue of NaN may arise due to numerical overflow when using FP16 for computation in these models(T5-Large).

[BUG] Incorrect Model Outputs When Using Beam Search

any update about num_beams > 1?

请问ARM架构支持吗，谢谢

没细看，直觉上改下那个汇编源文件应该就可以

‹
1
2