2404v6

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: 0.5939
• Accuracy: 0.8445
• Precision: 0.8451
• Recall: 0.8445
• F1: 0.8445
• Ratio: 0.4790

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: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 2
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Ratio
0.5731	0.2597	10	0.5646	0.8613	0.8645	0.8613	0.8610	0.4538
0.5515	0.5195	20	0.5569	0.8613	0.8626	0.8613	0.8612	0.4706
0.5164	0.7792	30	0.6079	0.8277	0.8363	0.8277	0.8266	0.5798
0.5641	1.0390	40	0.5728	0.8571	0.8608	0.8571	0.8568	0.4496
0.4665	1.2987	50	0.5992	0.8403	0.8407	0.8403	0.8403	0.5168
0.4632	1.5584	60	0.5990	0.8613	0.8634	0.8613	0.8611	0.4622
0.4456	1.8182	70	0.5939	0.8445	0.8451	0.8445	0.8445	0.4790

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1