VideoMAE_BdSLW60_FrameRate_Corrected_with_Augment_20_epoch_val_U5

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: 0.7120
• Accuracy: 0.855
• Precision: 0.8661
• Recall: 0.855
• F1: 0.8387

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: 18560
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.2576	0.0501	929	2.1251	0.5317	0.4875	0.5317	0.4654
0.5364	1.0501	1858	0.9989	0.7233	0.7522	0.7233	0.6834
0.1722	2.0501	2787	0.4298	0.8683	0.8966	0.8683	0.8524
0.123	3.0501	3716	0.3562	0.8983	0.9358	0.8983	0.8948
0.106	4.0501	4645	0.5168	0.875	0.9061	0.875	0.8704
0.0836	5.0501	5574	0.6956	0.86	0.8826	0.86	0.8485
0.0517	6.0501	6503	0.7006	0.8717	0.9056	0.8717	0.8545
0.0219	7.0501	7432	0.5776	0.8817	0.9109	0.8817	0.8711
0.0158	8.0501	8361	0.7120	0.855	0.8661	0.855	0.8387

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1