videomae-base-short-finetuned-kinetics-finetuned-rwf2000-epochs8-batch8-k

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

• Loss: 1.0855
• Accuracy: 0.6764

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
0.3699	0.06	200	0.2920	0.8925
0.3127	1.06	400	0.6035	0.77
0.2864	2.06	600	0.5088	0.8237
0.1739	3.06	800	0.6310	0.765
0.1835	4.06	1000	0.3643	0.8738
0.1075	5.06	1200	0.3455	0.8862
0.1362	6.06	1400	0.9145	0.75
0.2958	7.06	1600	0.3544	0.895
0.1571	8.06	1800	0.3748	0.8912
0.2448	9.06	2000	0.3646	0.8975
0.1939	10.06	2200	0.4430	0.8762
0.0666	11.06	2400	0.4916	0.8762
0.1958	12.06	2600	0.5114	0.8638
0.1063	13.06	2800	0.5701	0.8612
0.1047	14.06	3000	0.5226	0.8688
0.0131	15.06	3200	0.4656	0.8812

Framework versions

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