videomae-base-finetuned-kisa

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: 4.2647
• Accuracy: 0.4913

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: 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: 0.05
• training_steps: 2725

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.0741	0.0404	110	1.1353	0.5
0.0259	1.0404	220	3.7142	0.1183
0.5784	2.0404	330	2.2692	0.5
0.1384	3.0404	440	1.3726	0.5178
0.513	4.0404	550	2.5340	0.3728
0.0156	5.0404	660	2.3487	0.2041
0.0033	6.0404	770	4.4601	0.1953
0.0071	7.0404	880	4.6045	0.0917
0.004	8.0404	990	3.4062	0.4083
0.0017	9.0404	1100	2.4961	0.4941
0.4934	10.0404	1210	2.9785	0.4941
0.43	11.0404	1320	3.7030	0.5207
0.0014	12.0404	1430	2.5479	0.2012
0.0021	13.0404	1540	4.0235	0.3195
0.2387	14.0404	1650	4.6049	0.2337
0.0009	15.0404	1760	4.3070	0.2485
0.0004	16.0404	1870	4.4573	0.2515
0.5939	17.0404	1980	4.3423	0.3550
0.0013	18.0404	2090	4.3365	0.3047
0.0015	19.0404	2200	4.0964	0.2426
0.0032	20.0404	2310	4.1795	0.2988
0.0006	21.0404	2420	4.1612	0.3136

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0