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: 1.1144
• Accuracy: 0.5987

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: 16
• eval_batch_size: 16
• 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: 500

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.5857	0.04	22	1.5601	0.3459
1.5269	1.04	44	1.4939	0.3151
1.3172	2.04	66	1.1239	0.4658
1.2214	3.04	88	1.1980	0.4623
1.208	4.04	110	1.2465	0.4555
1.0422	5.04	132	1.2294	0.4966
1.0219	6.04	154	1.1516	0.5240
0.9113	7.04	176	1.2117	0.5068
1.035	8.04	198	1.0770	0.5616
0.8992	9.04	220	1.0658	0.5582
0.7292	10.04	242	1.2217	0.5445
0.8545	11.04	264	1.0260	0.5514
0.5855	12.04	286	1.0646	0.5993
0.7059	13.04	308	1.1769	0.5582
0.8109	14.04	330	1.1800	0.5137
0.6262	15.04	352	1.0740	0.5890
0.6297	16.04	374	1.0434	0.5719
0.7063	17.04	396	1.0205	0.5548
0.587	18.04	418	0.9799	0.6027
0.6087	19.04	440	0.9967	0.6164
0.5973	20.04	462	0.9683	0.6096
0.6971	21.04	484	1.0395	0.6027
0.6395	22.03	500	1.0663	0.5993

Framework versions

• Transformers 4.35.2
• Pytorch 1.13.1
• Datasets 2.15.0
• Tokenizers 0.15.0