videomae-base-finetuned-kinetics-allkisa-crop-background-0312-clip_duration-abnormal12_resize

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

• Loss: 0.3122
• Accuracy: 0.9393

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: 8
• eval_batch_size: 8
• seed: 42
• 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: 0.05
• training_steps: 29600

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6616	0.01	296	0.3786	0.8675
0.1816	1.01	592	0.5763	0.8352
0.0266	2.01	888	0.6628	0.8078
0.4168	3.01	1184	0.4009	0.8562
0.0271	4.01	1480	0.4466	0.8675
0.4371	5.01	1776	0.5005	0.8498
0.0269	6.01	2072	0.5684	0.8498
0.0061	7.01	2368	0.5858	0.8643
0.0262	8.01	2664	0.4571	0.8901
0.0003	9.01	2960	0.6348	0.8627
0.0004	10.01	3256	0.6531	0.8708
0.002	11.01	3552	0.5016	0.8821
0.0048	12.01	3848	0.6594	0.8578
0.5788	13.01	4144	0.5558	0.8934
0.0002	14.01	4440	0.6740	0.8611
0.0001	15.01	4736	0.5346	0.8950
0.007	16.01	5032	0.6110	0.8885
0.0001	17.01	5328	0.6141	0.8869
0.0363	18.01	5624	0.5504	0.8934
0.0002	19.01	5920	0.5372	0.8982
0.4009	20.01	6216	0.5422	0.8998
0.0049	21.01	6512	0.5687	0.8918
0.0048	22.01	6808	0.5355	0.8982
0.0	23.01	7104	0.5720	0.8982
0.0038	24.01	7400	0.5421	0.9047
0.0004	25.01	7696	0.6254	0.8950
0.0002	26.01	7992	0.5809	0.8998
0.0031	27.01	8288	0.5869	0.8966
0.0011	28.01	8584	0.6073	0.8950
0.0001	29.01	8880	0.5942	0.9015
0.5734	30.01	9176	0.5724	0.8998
0.0001	31.01	9472	0.5558	0.9047
0.0001	32.01	9768	0.5694	0.8982
0.0	33.01	10064	0.5874	0.9031
0.0	34.01	10360	0.6037	0.8966

Framework versions

• Transformers 4.48.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0