VideoMAE_kinetics__wlasl_2000_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: 7.7839
• Accuracy: 0.0041

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
30.6291	0.05	1786	7.6317	0.0008
30.5613	1.0500	3572	7.6099	0.0010
30.5034	2.0500	5358	7.6138	0.0008
30.3864	3.0500	7145	7.5879	0.0015
30.2681	4.05	8931	7.6050	0.0015
30.0755	5.0500	10717	7.6299	0.0028
29.8447	6.0500	12503	7.6543	0.0013
29.5841	7.0500	14290	7.6993	0.0015
29.3254	8.05	16076	7.7229	0.0020
29.0604	9.0500	17862	7.7460	0.0026
28.7758	10.0500	19648	7.7552	0.0033
28.4736	11.0500	21435	7.7740	0.0041
28.2064	12.05	23221	7.7849	0.0031
27.9399	13.0500	25007	7.7951	0.0036
27.6586	14.0500	26793	7.7796	0.0033
27.3957	15.0500	28580	7.7798	0.0038
27.1794	16.05	30366	7.7839	0.0041

Framework versions

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