adriansanz/te-zsc-synthetic_5ep_2805

080524_epoch_5

This model is a fine-tuned version of projecte-aina/roberta-base-ca-v2-cased-te on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3399
• Accuracy: 0.981
• Precision: 0.9810
• Recall: 0.981
• F1: 0.9810
• Ratio: 0.495

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: 10
• eval_batch_size: 2
• seed: 47
• gradient_accumulation_steps: 2
• total_train_batch_size: 20
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• lr_scheduler_warmup_steps: 4
• num_epochs: 1
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Ratio
0.3013	0.0333	10	0.3474	0.978	0.9783	0.978	0.9780	0.488
0.3087	0.0667	20	0.3471	0.979	0.9790	0.979	0.9790	0.495
0.3181	0.1	30	0.3527	0.975	0.9752	0.975	0.9750	0.489
0.3134	0.1333	40	0.3602	0.971	0.9714	0.971	0.9710	0.485
0.3002	0.1667	50	0.3481	0.979	0.9790	0.979	0.9790	0.501
0.3226	0.2	60	0.3547	0.978	0.9780	0.978	0.9780	0.496
0.2919	0.2333	70	0.3687	0.972	0.9724	0.972	0.9720	0.486
0.2932	0.2667	80	0.3822	0.965	0.9664	0.965	0.9650	0.473
0.3303	0.3	90	0.3754	0.969	0.9700	0.969	0.9690	0.477
0.3162	0.3333	100	0.3557	0.975	0.9750	0.975	0.9750	0.505
0.3012	0.3667	110	0.3554	0.974	0.9741	0.974	0.9740	0.506
0.3337	0.4	120	0.3629	0.972	0.9725	0.972	0.9720	0.484
0.3007	0.4333	130	0.3492	0.979	0.9792	0.979	0.9790	0.491
0.3283	0.4667	140	0.3467	0.979	0.9790	0.979	0.9790	0.495
0.3238	0.5	150	0.3410	0.981	0.9810	0.981	0.9810	0.497
0.3076	0.5333	160	0.3387	0.982	0.9820	0.982	0.9820	0.498
0.3348	0.5667	170	0.3375	0.982	0.9820	0.982	0.9820	0.498
0.3258	0.6	180	0.3401	0.98	0.9801	0.98	0.9800	0.494
0.3195	0.6333	190	0.3424	0.978	0.9781	0.978	0.9780	0.492
0.31	0.6667	200	0.3392	0.981	0.9810	0.981	0.9810	0.495
0.3407	0.7	210	0.3393	0.982	0.9820	0.982	0.9820	0.502
0.3494	0.7333	220	0.3413	0.981	0.9810	0.981	0.9810	0.501
0.3574	0.7667	230	0.3402	0.982	0.9820	0.982	0.9820	0.496
0.3379	0.8	240	0.3385	0.982	0.9820	0.982	0.9820	0.496
0.3532	0.8333	250	0.3385	0.982	0.9820	0.982	0.9820	0.496
0.318	0.8667	260	0.3425	0.98	0.9801	0.98	0.9800	0.494
0.3475	0.9	270	0.3432	0.98	0.9801	0.98	0.9800	0.494
0.3142	0.9333	280	0.3408	0.981	0.9810	0.981	0.9810	0.495
0.3421	0.9667	290	0.3404	0.981	0.9810	0.981	0.9810	0.495
0.2935	1.0	300	0.3399	0.981	0.9810	0.981	0.9810	0.495

Framework versions

• Transformers 4.41.0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1