videomae-base-finetuned-good-gestureUnitV13

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: 0.9540
• Loss: 0.1660
• Accuracy Gunit: 0.9048
• 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: 320

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Accuracy Gunit	Accuracy Nothing
0.6625	0.0531	17	0.5893	0.6896	1.0	0.2581
0.6423	1.0531	34	0.5893	0.6737	1.0	0.2581
0.6252	2.0531	51	0.6071	0.6100	0.72	0.5161
0.5455	3.0531	68	0.7321	0.5558	0.96	0.5484
0.4912	4.0531	85	0.8393	0.4599	0.76	0.9032
0.383	5.0531	102	0.8571	0.3426	0.72	0.9677
0.2696	6.0531	119	0.8393	0.3441	0.72	0.9355
0.1547	7.0531	136	0.8571	0.3526	0.72	0.9677
0.3174	8.0531	153	0.8929	0.3234	0.8	0.9677
0.1324	9.0531	170	0.8393	0.5808	0.68	0.9677
0.0265	10.0531	187	0.8393	0.6995	0.68	0.9677
0.2346	11.0531	204	0.8571	0.7486	0.68	1.0
0.0597	12.0531	221	0.8393	0.7536	0.68	0.9677
0.0103	13.0531	238	0.8929	0.4090	0.88	0.9032
0.0358	14.0531	255	0.8571	0.8257	0.68	1.0
0.0047	15.0531	272	0.875	0.6291	0.76	0.9677
0.0054	16.0531	289	0.8393	0.8420	0.68	0.9677
0.0024	17.0531	306	0.8393	0.8389	0.68	0.9677
0.0027	18.0437	320	0.8393	0.7864	0.68	0.9677

Framework versions

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