VideoMAE_base__wlasl_100_20epoch

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: 3.6747
• Accuracy: 0.1331

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
18.6651	0.05	180	4.6356	0.0059
18.6425	1.0499	360	4.6281	0.0178
18.5724	2.0499	540	4.6207	0.0178
18.384	3.0501	721	4.6615	0.0148
18.4347	4.05	901	4.6206	0.0148
18.3331	5.0499	1081	4.6171	0.0266
18.1997	6.0499	1261	4.6364	0.0296
17.8738	7.0501	1442	4.6173	0.0237
17.6366	8.05	1622	4.4824	0.0237
16.9514	9.0499	1802	4.3944	0.0385
16.3058	10.0499	1982	4.2680	0.0414
15.6024	11.0501	2163	4.2257	0.0414
14.9672	12.05	2343	4.1320	0.0621
14.1623	13.0499	2523	4.0324	0.0503
13.3095	14.0499	2703	3.9649	0.0888
12.3516	15.0501	2884	3.8141	0.1183
11.5769	16.05	3064	3.7283	0.1183
10.8459	17.0499	3244	3.7133	0.1331
10.2769	18.0499	3424	3.6848	0.1450
9.8166	19.0487	3600	3.6747	0.1331

Framework versions

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