Shanyong Yu comments

Results 17 comments of


                                            Shanyong Yu

no_sync equivalent used for gradient accumulation

@amsword Have you solved this problem？

Methods to solve or mitigate the AMP training (inference) failing problem

Hi @qijiaxing can you show the results of these two experimental Baseline? thanks.

[Help wanted] Customize onnxruntime

Hi @w11wo, in the end, how small did manage to reduce the size of onnxruntime static lib? (ios-arm64) I try to reduce it to 20M+, it seems to be the...

Need help in fine-tuning whisper

@Naramkeshav59 Did you get any resolution on this? I am facing the same issues

runtime iOS Demo cannot compile successfully

@LauraGPT @andyweiqiu I also encountered the same problem, can you help us?

runtime iOS Demo cannot compile successfully

@summerHearts 自己重写一个吧😂

API参数没注释

好的，谢谢！

Rockchip RK3588上npu推理耗时比cpu长

我看cpu利用率是20%左右，npu 小于10%

Rockchip RK3588上npu推理耗时比cpu长

用cpu跑，cpu利用率差不多也是20%多

Rockchip RK3588上npu推理耗时比cpu长

统一用sherpa-onnx 用npu跑，cpu利用率最高70%多：用cpu跑，cpu利用率最高100%多：