videomae-base-finetuned-ucf101-subset

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: 0.3320
• Accuracy: 0.9290

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 600
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.3375	0.0617	37	2.1297	0.3
1.7263	1.0625	75	1.4907	0.4714
0.675	2.0617	112	0.6070	0.8714
0.3702	3.0625	150	0.4089	0.8571
0.205	4.0617	187	0.4285	0.8286
0.3209	5.0625	225	0.2749	0.8714
0.1253	6.0617	262	0.0571	0.9857
0.1052	7.0625	300	0.2550	0.9429
0.1586	8.0617	337	0.1588	0.9429
0.0498	9.0625	375	0.0736	0.9857
0.0958	10.0617	412	0.0911	0.9571
0.1095	11.0625	450	0.0448	0.9857
0.0065	12.0617	487	0.0734	0.9857
0.0019	13.0625	525	0.0603	0.9857
0.0018	14.0617	562	0.0607	0.9857
0.0018	15.0625	600	0.0634	0.9857

Framework versions

• Transformers 4.46.3
• Pytorch 2.0.1+cu117
• Datasets 3.2.0
• Tokenizers 0.20.3