videomae-base-finetuned-good-gestureUnitsV2

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:

• Accuracy: 0.8932
• Loss: 0.3541
• Accuracy Gunit: 0.5417
• Accuracy Nothing: 1.0

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 170

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Accuracy Gunit	Accuracy Nothing
0.7018	0.1	17	0.2810	0.7374	0.75	0.0494
0.6812	1.1	34	0.3223	0.7290	0.925	0.0247
0.606	2.1	51	0.5537	0.6634	0.85	0.4074
0.604	3.1	68	0.4298	0.6624	0.95	0.1728
0.5523	4.1	85	0.8512	0.5508	0.85	0.8519
0.4609	5.1	102	0.8595	0.4713	0.875	0.8519
0.4126	6.1	119	0.9421	0.3119	0.85	0.9877
0.2594	7.1	136	0.9421	0.2562	0.85	0.9877
0.1955	8.1	153	0.9174	0.2955	0.875	0.9383
0.2203	9.1	170	0.9421	0.2354	0.85	0.9877

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1