VideoMAE_BdSLW60_FrameRateCorrected_withAug_100

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: 0.5588
• Accuracy: 0.8973

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: 22400

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.0699	0.0400	897	2.4541	0.4753
0.6366	1.0401	1795	0.6832	0.84
0.2253	2.0401	2693	0.3464	0.9024
0.1229	3.0401	3591	0.1467	0.9647
0.1045	4.0400	4488	0.1459	0.9635
0.0631	5.0401	5386	0.1313	0.9718
0.0736	6.0401	6284	0.1807	0.9635
0.0673	7.0401	7182	0.1464	0.9694
0.0239	8.0400	8079	0.1932	0.9576
0.0868	9.0401	8977	0.0563	0.9882
0.0016	10.0401	9875	0.0844	0.9776
0.0318	11.0401	10773	0.1123	0.9753
0.0144	12.0400	11670	0.0499	0.9894
0.0028	13.0401	12568	0.0809	0.9871
0.0074	14.0401	13466	0.0455	0.9929
0.0002	15.0401	14364	0.0581	0.9906
0.0077	16.0400	15261	0.0502	0.9894
0.0005	17.0401	16159	0.0407	0.9929
0.0004	18.0401	17057	0.0550	0.9906
0.0001	19.0401	17955	0.0583	0.9929

Framework versions

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