videomae-base-finetuned-rwf2000-subset___v4

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: 0.2571
• Accuracy: 0.91

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: 5.5e-05
• train_batch_size: 16
• eval_batch_size: 16
• 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: 0.1
• training_steps: 1500

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5186	0.0667	100	0.4852	0.7525
0.4113	1.0667	200	0.8493	0.6512
0.3743	2.0667	300	0.8014	0.6675
0.2878	3.0667	400	0.4332	0.83
0.2419	4.0667	500	0.4650	0.8225
0.2565	5.0667	600	0.6123	0.7638
0.3317	6.0667	700	0.5332	0.7725
0.2739	7.0667	800	0.4160	0.82
0.1534	8.0667	900	0.4775	0.82
0.2573	9.0667	1000	0.4197	0.8413
0.16	10.0667	1100	0.4305	0.8413
0.1661	11.0667	1200	0.6462	0.7913
0.1194	12.0667	1300	0.7474	0.7925
0.1787	13.0667	1400	0.6112	0.8363
0.1953	14.0667	1500	0.6876	0.8013

Framework versions

• Transformers 4.46.2
• Pytorch 1.13.1+cu117
• Datasets 3.1.0
• Tokenizers 0.20.3