videomae-large-cctv-brawl_extended_v2

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

• Loss: 0.7025
• Accuracy: 0.8993

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-07
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 37695

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4212	0.07	2514	0.4737	0.7236
0.2813	1.07	5028	0.3922	0.8350
0.1918	2.07	7542	0.5665	0.8734
0.2002	3.07	10056	0.6058	0.8808
0.1921	4.07	12570	0.6136	0.8867
0.1713	5.07	15084	0.6446	0.8914
0.006	6.07	17598	0.6783	0.8918
0.004	7.07	20112	0.7586	0.8863
0.201	8.07	22626	0.7732	0.8906
0.0237	9.07	25140	0.7025	0.8993
0.0003	10.07	27654	0.7510	0.8957
0.1943	11.07	30168	0.7849	0.8989
0.0001	12.07	32682	0.8075	0.8981
0.0	13.07	35196	0.8347	0.8961
0.0001	14.07	37695	0.8329	0.8985

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1