videomae-base-finetuned-kinetics-finetuned-rwf2000-epochs8-batch8-kb

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

• Loss: 0.5482
• Accuracy: 0.7298

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

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4492	0.06	200	0.2361	0.905
0.3517	1.06	400	0.5648	0.8137
0.2255	2.06	600	0.7592	0.7575
0.1983	3.06	800	0.4803	0.835
0.2305	4.06	1000	0.4290	0.8738
0.1276	5.06	1200	0.4317	0.8762
0.1597	6.06	1400	1.3708	0.6937
0.3088	7.06	1600	0.3974	0.8862
0.2687	8.06	1800	0.5986	0.85
0.2085	9.06	2000	0.4264	0.8862
0.1338	10.06	2200	0.5015	0.8675
0.2191	11.06	2400	0.7103	0.845
0.2255	12.06	2600	0.4939	0.8762
0.0298	13.06	2800	0.6338	0.8612
0.0687	14.06	3000	0.5350	0.8738
0.0146	15.06	3200	0.4770	0.8838

Framework versions

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