VideoMAE_base_wlasl_100__signer_20ep_coR

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: 3.1409
• Accuracy: 0.2367

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: 3600
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
18.6682	0.05	180	4.6265	0.0178
18.6293	1.0499	360	4.6238	0.0178
18.5543	2.0499	540	4.6135	0.0178
18.3007	3.0501	721	4.6231	0.0237
18.3088	4.05	901	4.5822	0.0237
18.0334	5.0499	1081	4.5451	0.0207
17.5398	6.0499	1261	4.4164	0.0207
16.6553	7.0501	1442	4.2852	0.0325
16.0071	8.05	1622	4.2631	0.0414
15.2212	9.0499	1802	4.0466	0.0621
14.1984	10.0499	1982	3.9305	0.0799
13.0441	11.0501	2163	3.8182	0.1095
11.9689	12.05	2343	3.5890	0.1272
10.7357	13.0499	2523	3.4955	0.1598
9.6829	14.0499	2703	3.3952	0.1834
8.6462	15.0501	2884	3.3082	0.1746
7.7971	16.05	3064	3.2246	0.2160
7.0978	17.0499	3244	3.2038	0.2308
6.483	18.0499	3424	3.1605	0.2396
6.0651	19.0487	3600	3.1409	0.2367

Framework versions

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