videomae-base-finetuned-good-gesturePhaseV10

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: 0.9253
• Loss: 0.3122
• Accuracy Hold: 1.0
• Accuracy Stroke: 0.4286
• Accuracy Recovery: 0.7895
• Accuracy Preparation: 1.0
• Accuracy Unknown: 0.6429

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: 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: 630

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Accuracy Hold	Accuracy Stroke	Accuracy Recovery	Accuracy Preparation	Accuracy Unknown
1.1344	0.2016	127	0.6900	1.0021	0.0	0.0	0.0	1.0	0.0
0.5961	1.2016	254	0.7948	0.6022	0.2692	0.0	0.0588	0.9873	0.8182
0.3453	2.2016	381	0.8777	0.3925	0.8077	0.0	0.4118	0.9747	0.8636
0.1551	3.2016	508	0.9432	0.2178	0.9615	0.1667	0.7059	0.9937	0.9545
0.1213	4.1937	630	0.9476	0.2032	0.9615	0.6667	0.7059	1.0	0.8182

Framework versions

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