videomae-base-finetuned-good-gestureUnitsV3

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: 0.3490
• 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: 5e-06
• 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: 340

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Accuracy Gunit	Accuracy Nothing
0.7047	0.05	17	0.5620	0.7024	0.575	0.5556
0.6958	1.05	34	0.3802	0.7165	0.725	0.2099
0.6534	2.05	51	0.3719	0.7150	0.85	0.1358
0.655	3.05	68	0.3223	0.7506	0.975	0.0
0.6545	4.05	85	0.4298	0.6991	0.95	0.1728
0.6321	5.05	102	0.4711	0.6750	0.95	0.2346
0.5883	6.05	119	0.7438	0.6142	0.9	0.6667
0.5045	7.05	136	0.8182	0.5495	0.9	0.7778
0.455	8.05	153	0.7190	0.5723	0.925	0.6173
0.4191	9.05	170	0.9339	0.3709	0.85	0.9753
0.365	10.05	187	0.9091	0.3547	0.85	0.9383
0.2593	11.05	204	0.8926	0.3632	0.85	0.9136
0.2225	12.05	221	0.9421	0.2457	0.85	0.9877
0.2121	13.05	238	0.9256	0.2619	0.85	0.9630
0.1506	14.05	255	0.9504	0.2190	0.85	1.0
0.1335	15.05	272	0.9339	0.2221	0.85	0.9753
0.1039	16.05	289	0.9421	0.2107	0.85	0.9877
0.1036	17.05	306	0.9421	0.2059	0.85	0.9877
0.1019	18.05	323	0.9421	0.2063	0.85	0.9877
0.1181	19.05	340	0.9504	0.2103	0.85	1.0

Framework versions

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