Evaluation Results are not Consistent in Consecutive Evaluations & Sensitivty to Batch Size

Thank you for sharing this wonderful work! Could you help to look into the following two issues:

1. I tested the code on the HMDB51 dataset, the results can be inconsistent for two consecutive evaluations (run the LINE414 of main_videp.py test_stats = evaluate(data_loader_val, model, device) twice).

2. For fine-tuning with Swin Transformer, I ran the code with a smaller batch size (i.e., 32) on 4 3090 GPUs several times, and the results for tunning the linear layer are around 71+%. Is the large batch size making such a difference from the reported results of 74%?

Thank you very much in advance!

[ShoufaChen]

I see. For the evaluation between two training epochs, random selection of video frames exists:(https://github.com/ShoufaChen/AdaptFormer/blob/6967d676c1a5e5a11be2e2768a6e5c604bb043ed/datasets/kinetics.py#L286

The final results are obtained from https://github.com/ShoufaChen/AdaptFormer/blob/main/main_video.py#L436.

[bruceyo]

I see. For the evaluation between two training epochs, random selection of video frames exists:(

https://github.com/ShoufaChen/AdaptFormer/blob/6967d676c1a5e5a11be2e2768a6e5c604bb043ed/datasets/kinetics.py#L286

The final results are obtained from https://github.com/ShoufaChen/AdaptFormer/blob/main/main_video.py#L436.

Thank you:)