videomae-base-short-finetuned-ssv2-finetuned-rwf2000-epochs8-batch8-fp16

This model is a fine-tuned version of MCG-NJU/videomae-base-short-finetuned-ssv2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4339
• Accuracy: 0.4643

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 3200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4239	0.06	200	0.3879	0.82
0.4179	1.06	400	1.1635	0.6162
0.4329	2.06	600	0.8215	0.63
0.3051	3.06	800	0.5541	0.7412
0.172	4.06	1000	0.4696	0.8363
0.1955	5.06	1200	0.5384	0.78
0.2301	6.06	1400	1.3358	0.635
0.2995	7.06	1600	1.0372	0.7087
0.3789	8.06	1800	0.8670	0.7412
0.2525	9.06	2000	0.5886	0.8225
0.1846	10.06	2200	0.7851	0.735
0.1547	11.06	2400	0.8905	0.7638
0.2501	12.06	2600	0.9807	0.76
0.1046	13.06	2800	1.0419	0.7438
0.0786	14.06	3000	1.0128	0.7538
0.0178	15.06	3200	1.0156	0.75

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu117
• Datasets 2.8.0
• Tokenizers 0.13.2