videomae-base-finetuned-ucf101-subset

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: 0.3235
• Accuracy: 0.9355

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: 4
• eval_batch_size: 4
• 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: 1875

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.198	0.04	75	2.0868	0.4
1.3004	1.04	150	1.0574	0.6714
0.6807	2.04	225	0.5173	0.8
0.8481	3.04	300	0.9220	0.7
0.2129	4.04	375	0.4430	0.7857
0.3205	5.04	450	0.4711	0.8571
0.1059	6.04	525	0.0504	0.9857
0.0048	7.04	600	0.6079	0.8857
0.043	8.04	675	0.5023	0.8714
0.0061	9.04	750	0.5091	0.8571
0.0029	10.04	825	0.0924	0.9571
0.0022	11.04	900	0.0082	1.0
0.1303	12.04	975	0.2828	0.9429
0.0054	13.04	1050	0.1975	0.9429
0.0019	14.04	1125	0.1887	0.9429
0.0015	15.04	1200	0.7790	0.8571
0.0017	16.04	1275	0.2677	0.9286
0.0014	17.04	1350	0.3186	0.9286
0.0012	18.04	1425	0.2961	0.9429
0.0012	19.04	1500	0.4832	0.9
0.0011	20.04	1575	0.3789	0.9429
0.001	21.04	1650	0.2051	0.9571
0.0011	22.04	1725	0.1955	0.9714
0.0011	23.04	1800	0.1916	0.9714
0.0013	24.04	1875	0.1907	0.9714

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1