videberta-base_v1

This model is a fine-tuned version of Fsoft-AIC/videberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4110
• Accuracy: 0.8882
• Precision Macro: 0.7636
• Recall Macro: 0.7197
• F1 Macro: 0.7363
• F1 Weighted: 0.8843

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: 3e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• 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
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
0.8764	1.0	90	0.7142	0.6974	0.4684	0.4901	0.4759	0.6809
0.682	2.0	180	0.5610	0.7701	0.5261	0.5431	0.5253	0.7514
0.5221	3.0	270	0.4966	0.8294	0.5546	0.5817	0.5660	0.8102
0.429	4.0	360	0.4697	0.8395	0.6756	0.5881	0.5807	0.8204
0.3652	5.0	450	0.4085	0.8642	0.7889	0.6334	0.6442	0.8503
0.3638	6.0	540	0.4011	0.8743	0.8328	0.6359	0.6447	0.8591
0.3148	7.0	630	0.3770	0.8806	0.8160	0.6770	0.7037	0.8712
0.2928	8.0	720	0.3874	0.8825	0.8480	0.6751	0.7020	0.8724
0.2705	9.0	810	0.3800	0.8793	0.7808	0.7026	0.7254	0.8737
0.2397	10.0	900	0.3699	0.8882	0.8000	0.6991	0.7257	0.8810
0.2325	11.0	990	0.3837	0.8863	0.8213	0.6647	0.6855	0.8745
0.2158	12.0	1080	0.3721	0.8857	0.7843	0.7061	0.7296	0.8798
0.1985	13.0	1170	0.3878	0.8907	0.8037	0.7090	0.7362	0.8844
0.2035	14.0	1260	0.3784	0.8857	0.7685	0.7173	0.7363	0.8815
0.1805	15.0	1350	0.4019	0.8850	0.7565	0.7005	0.7193	0.8795
0.1808	16.0	1440	0.4085	0.8882	0.7732	0.7114	0.7322	0.8831
0.1646	17.0	1530	0.3906	0.8831	0.7496	0.7368	0.7427	0.8819
0.1687	18.0	1620	0.3998	0.8857	0.7606	0.7306	0.7431	0.8831
0.1636	19.0	1710	0.4107	0.8863	0.7594	0.7184	0.7341	0.8826
0.1634	20.0	1800	0.4110	0.8882	0.7636	0.7197	0.7363	0.8843

Framework versions

• Transformers 4.55.0
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4