VideoMAE_Base_wlasl_2000_longtail_20

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: 7.8221
• Accuracy: 0.0033

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
30.6409	0.05	1786	7.6310	0.0005
30.5597	1.0500	3572	7.6175	0.0005
30.4316	2.0500	5358	7.6035	0.0010
30.2683	3.0500	7145	7.5938	0.0020
30.0727	4.05	8931	7.6268	0.0018
29.84	5.0500	10717	7.6477	0.0026
29.5721	6.0500	12503	7.6825	0.0023
29.2352	7.0500	14290	7.7271	0.0023
28.9425	8.05	16076	7.7662	0.0041
28.6146	9.0500	17862	7.7746	0.0031
28.3135	10.0500	19648	7.7994	0.0028
27.985	11.0500	21435	7.8092	0.0036
27.6736	12.05	23221	7.8222	0.0028
27.3741	13.0500	25007	7.8221	0.0033

Framework versions

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