videomae_v1_rwf-2000

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: 0.5974
• Accuracy: 0.845
• F1: 0.8447
• Precision: 0.8475

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: 2e-05
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 18
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 2024

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision
1.4589	0.0415	84	1.4350	0.7	0.6992	0.7020
0.7575	1.0417	169	0.8168	0.7812	0.7738	0.8238
0.6272	2.0418	254	0.5022	0.875	0.8740	0.8868
0.4146	3.0415	338	0.4904	0.8688	0.8686	0.8702
0.2696	4.0417	423	0.4742	0.85	0.85	0.85
0.3349	5.0418	508	0.4333	0.875	0.8748	0.8771

Framework versions

• Transformers 4.44.0
• Pytorch 2.0.1+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1