videomae-large-fight_22-01-2024

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: 0.6263
• Accuracy: 0.8565
• Precision: 0.8502
• Recall: 0.8655

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: 9080

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall
0.6582	0.05	454	0.6970	0.5695	0.5660	0.5964
0.6712	1.05	908	0.6281	0.6390	0.6202	0.7175
0.5664	2.05	1362	0.6718	0.6457	0.6555	0.6143
0.5645	3.05	1816	0.5835	0.7018	0.6974	0.7130
0.4259	4.05	2270	0.5497	0.7197	0.7402	0.6771
0.3542	5.05	2724	0.5509	0.7466	0.7434	0.7534
0.3676	6.05	3178	0.4956	0.7623	0.7532	0.7803
0.2656	7.05	3632	0.5263	0.7534	0.7811	0.7040
0.4675	8.05	4086	0.5216	0.7915	0.8009	0.7758
0.1434	9.05	4540	0.4744	0.8094	0.8136	0.8027
0.1389	10.05	4994	0.5389	0.8318	0.8274	0.8386
0.3228	11.05	5448	0.5345	0.8341	0.8599	0.7982
0.1044	12.05	5902	0.5729	0.8341	0.8465	0.8161
0.0305	13.05	6356	0.5812	0.8363	0.8378	0.8341
0.1256	14.05	6810	0.5806	0.8520	0.8489	0.8565
0.2735	15.05	7264	0.5713	0.8520	0.8618	0.8386
0.2376	16.05	7718	0.6030	0.8498	0.8578	0.8386
0.2978	17.05	8172	0.6263	0.8565	0.8502	0.8655
0.3872	18.05	8626	0.6099	0.8520	0.8489	0.8565
0.6629	19.05	9080	0.6142	0.8543	0.8496	0.8610

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0