VideoMAE_base_wlasl100_20epoch_Signers

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: 3.5389
• Accuracy: 0.1864

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
18.6481	0.05	180	4.6265	0.0207
18.6251	1.0499	360	4.6404	0.0266
18.5402	2.0499	540	4.6303	0.0059
18.3373	3.0501	721	4.6409	0.0237
18.3156	4.05	901	4.6281	0.0118
18.1158	5.0499	1081	4.5800	0.0296
17.8542	6.0499	1261	4.5729	0.0237
17.3491	7.0501	1442	4.4994	0.0296
16.7036	8.05	1622	4.3456	0.0355
15.9584	9.0499	1802	4.2656	0.0355
15.2418	10.0499	1982	4.1895	0.0355
14.3282	11.0501	2163	4.1120	0.0562
13.4384	12.05	2343	4.0232	0.0533
12.3649	13.0499	2523	3.8854	0.1006
11.3044	14.0499	2703	3.7767	0.1095
10.2946	15.0501	2884	3.7317	0.1420
9.5039	16.05	3064	3.6387	0.1746
8.7162	17.0499	3244	3.5911	0.1805
8.1393	18.0499	3424	3.5438	0.1982
7.716	19.0487	3600	3.5389	0.1864

Framework versions

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