videomae_wlasl_100_30ep_coR

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: 0.9268
• Accuracy: 0.7544

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: 5400
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
18.4807	0.0333	180	4.5696	0.0296
17.3806	1.0333	360	3.9577	0.1302
13.6869	2.0332	540	3.1247	0.3846
9.4867	3.0334	721	2.3584	0.5385
6.2563	4.0333	901	1.8838	0.6302
3.9622	5.0333	1081	1.5797	0.6686
2.4851	6.0332	1261	1.4008	0.6746
1.4884	7.0334	1442	1.2604	0.7071
0.8963	8.0333	1622	1.1644	0.7249
0.5257	9.0333	1802	1.1132	0.7249
0.3328	10.0332	1982	1.1301	0.7101
0.1728	11.0334	2163	0.9261	0.7544
0.1272	12.0333	2343	0.9679	0.7515
0.0949	13.0333	2523	0.9419	0.7485
0.0832	14.0332	2703	0.9366	0.7426
0.0671	15.0334	2884	0.9128	0.7722
0.0563	16.0333	3064	0.8947	0.7663
0.0656	17.0333	3244	0.8782	0.7574
0.0506	18.0332	3424	0.8957	0.7722
0.0555	19.0334	3605	0.8948	0.7781
0.0765	20.0333	3785	0.9053	0.7722
0.0616	21.0333	3965	0.9024	0.7633
0.0359	22.0332	4145	0.9499	0.7574
0.0568	23.0334	4326	0.9315	0.7515
0.0562	24.0333	4506	0.9268	0.7544

Framework versions

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