videomae-base-finetuned-kinetics-0416_final_5sec_org_balanced_replacecroppretest

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

• Loss: 0.9630
• Accuracy: 0.8330

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch 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: 0.05
• training_steps: 57800

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.0448	0.0100	579	0.6481	0.8053
0.0015	1.0100	1158	0.7120	0.8430
0.0019	2.0100	1737	0.8507	0.8439
1.0975	3.0100	2316	0.7177	0.8412
0.0061	4.0100	2895	0.4413	0.8842
0.0047	5.0100	3474	0.8424	0.8246
0.001	6.0100	4053	0.6867	0.8807
0.0001	7.0100	4632	0.6921	0.8895
0.0012	8.0100	5211	0.9348	0.8518
0.0001	9.0100	5790	1.2715	0.8175
0.0076	10.0100	6369	0.9247	0.8544
0.0088	11.0100	6948	1.1050	0.8404
0.0	12.0100	7527	0.7281	0.8772
0.0001	13.0100	8106	0.8381	0.8763
0.0026	14.0100	8685	0.7997	0.8772
0.0099	15.0100	9264	0.9333	0.8649
0.0	16.0100	9843	1.0398	0.8430
0.0001	17.0100	10422	0.8570	0.8719
0.0012	18.0100	11001	0.9378	0.8649
0.0002	19.0100	11580	1.0377	0.8553
0.0004	20.0100	12159	0.8902	0.8684
0.0	21.0100	12738	1.2577	0.8377
0.2182	22.0100	13317	1.1700	0.8509

Framework versions

• Transformers 4.48.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0