Hal_videberta-base_finetuned

This model is a fine-tuned version of Fsoft-AIC/videberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8824
• Accuracy: 0.6907
• F1: 0.6907

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.104	1.0	350	1.0995	0.3493	0.3493
1.1055	2.0	700	1.0982	0.3493	0.3493
1.0983	3.0	1050	1.0992	0.35	0.35
1.0974	4.0	1400	1.0983	0.35	0.35
1.0959	5.0	1750	1.0954	0.3507	0.3507
0.834	6.0	2100	0.8176	0.6471	0.6471
0.7454	7.0	2450	0.8176	0.6571	0.6571
0.7321	8.0	2800	0.7939	0.6693	0.6693
0.6003	9.0	3150	0.8458	0.6657	0.6657
0.602	10.0	3500	0.8261	0.6821	0.6821
0.4936	11.0	3850	0.9280	0.6629	0.6629
0.4742	12.0	4200	0.8824	0.6907	0.6907
0.452	13.0	4550	1.0194	0.6714	0.6714
0.3957	14.0	4900	1.0482	0.6657	0.6657
0.4032	15.0	5250	1.0783	0.6686	0.6686

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0