VideoMAE_wlasl_2000_20_epochs

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: 2.4885
• Accuracy: 0.4484

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 35720
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
30.4463	0.05	1786	7.5937	0.0008
29.0015	1.0500	3572	6.7471	0.0209
24.748	2.0500	5358	5.8146	0.0855
20.3678	3.0500	7145	4.9957	0.1599
16.2208	4.05	8931	4.2713	0.2326
12.4666	5.0500	10717	3.7028	0.3018
9.2225	6.0500	12503	3.2447	0.3547
6.6389	7.0500	14290	2.9646	0.3820
4.7134	8.05	16076	2.7632	0.4027
3.2725	9.0500	17862	2.6250	0.4203
2.309	10.0500	19648	2.5632	0.4175
1.677	11.0500	21435	2.5348	0.4206
1.2664	12.05	23221	2.4956	0.4300
1.0	13.0500	25007	2.4907	0.4262
0.8149	14.0500	26793	2.4883	0.4354
0.6907	15.0500	28580	2.4774	0.4428
0.5841	16.05	30366	2.4884	0.4428
0.533	17.0500	32152	2.4878	0.4469
0.4575	18.0500	33938	2.4865	0.4469
0.3969	19.0499	35720	2.4885	0.4484

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.1