videomae-finetuned-v2

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: 0.4511
• Accuracy: 0.825

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: 16
• eval_batch_size: 16
• 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: 90

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.0556	5	0.6644	0.52
0.649	1.0556	10	0.5890	0.68
0.649	2.0556	15	0.6915	0.48
0.5368	3.0556	20	0.5428	0.48
0.5368	4.0556	25	0.2507	0.92
0.3519	5.0556	30	0.6503	0.68
0.3519	6.0556	35	0.6544	0.68
0.4579	7.0556	40	0.2332	0.92
0.4579	8.0556	45	0.4506	0.88
0.3166	9.0556	50	0.2587	0.88
0.3166	10.0556	55	0.1353	0.92
0.2761	11.0556	60	0.3067	0.92
0.2761	12.0556	65	0.4782	0.84
0.2316	13.0556	70	0.3868	0.84
0.2316	14.0556	75	0.3565	0.88
0.173	15.0556	80	0.3623	0.92
0.173	16.0556	85	0.2870	0.92
0.1488	17.0556	90	0.2836	0.92

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1