ner-roberta-large-lenerbr

This model is a fine-tuned version of roberta-large on the lener_br dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Precision: 0.8310
• Recall: 0.8987
• F1: 0.8635
• Accuracy: 0.9723

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: 4
• eval_batch_size: 8
• 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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0972	1.0	1957	nan	0.7404	0.8191	0.7778	0.9534
0.0712	2.0	3914	nan	0.7964	0.8437	0.8194	0.9584
0.0477	3.0	5871	nan	0.7845	0.8803	0.8296	0.9650
0.0243	4.0	7828	nan	0.7938	0.8664	0.8285	0.9684
0.0244	5.0	9785	nan	0.7611	0.9106	0.8291	0.9664
0.0322	6.0	11742	nan	0.7793	0.8921	0.8319	0.9672
0.0132	7.0	13699	nan	0.8310	0.8987	0.8635	0.9723
0.0156	8.0	15656	nan	0.7429	0.9170	0.8208	0.9656
0.0082	9.0	17613	nan	0.7658	0.9082	0.8309	0.9668
0.0032	10.0	19570	nan	0.7819	0.9095	0.8409	0.9697

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1