videomae-base-finetuned-ucf101-subset

This model is a fine-tuned version of MCG-NJU/videomae-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0279
• Accuracy: 0.9949

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 38500

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1974	0.0100	386	1.1477	0.4549
0.3532	1.0100	772	0.4044	0.8611
0.2043	2.0100	1158	0.3527	0.9010
0.1896	3.0100	1544	0.2104	0.9288
0.2453	4.0100	1930	0.2778	0.9340
0.0189	5.0100	2316	0.0706	0.9774
0.0443	6.0100	2702	0.0838	0.9792
0.0673	7.0100	3088	0.0385	0.9896
0.0009	8.0100	3474	0.0369	0.9913
0.1916	9.0100	3860	0.0349	0.9913
0.0569	10.0100	4246	0.1056	0.9774
0.0014	11.0100	4632	0.0046	0.9965
0.2	12.0100	5018	0.0107	0.9965
0.0706	13.0100	5404	0.0872	0.9844
0.0927	14.0100	5790	0.1275	0.9688
0.0012	15.0100	6176	0.0815	0.9774
0.0944	16.0100	6562	0.0228	0.9913
0.0003	17.0100	6948	0.0336	0.9878
0.0021	18.0100	7334	0.0166	0.9948
0.0002	19.0100	7720	0.0209	0.9965
0.052	20.0100	8106	0.0435	0.9896
0.0002	21.0100	8492	0.0155	0.9983

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.1+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1