QingshuiL

Results 5 comments of QingshuiL

那确实很不错了。请问是在哪个数据上测的,eval脚本是不是没有放出来。

感谢博主的回复。博主有了解过在中文识别领域有哪些比较权威的 benchmark 吗?

[single_modality](https://github.com/OpenGVLab/InternVideo/tree/main/InternVideo2/single_modality/run_finetuning.py) 我在 run_finetuning.py 脚本下第697行看到了评估数据精度的函数 final_test() ,这应该就是我在找的 。

I can only use the a100 GPU now. Will the quantization and inference of the A100 GPU be supported in the future? How many A100-40G GPUs are required if supported?

You can try downgrading the transformers == 4.51.2.