xlm-roberta-large-finetuned-ner-vlsp2021-3090-1July-1

This model is a fine-tuned version of FacebookAI/xlm-roberta-large on the None dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.1332
• eval_ATETIME: {'precision': 0.8748768472906404, 'recall': 0.8862275449101796, 'f1': 0.8805156172533465, 'number': 1002}
• eval_DDRESS: {'precision': 0.7837837837837838, 'recall': 1.0, 'f1': 0.8787878787878788, 'number': 29}
• eval_ERSON: {'precision': 0.9496365524402908, 'recall': 0.9631384939441812, 'f1': 0.9563398692810458, 'number': 1899}
• eval_ERSONTYPE: {'precision': 0.7142857142857143, 'recall': 0.7602339181286549, 'f1': 0.7365439093484419, 'number': 684}
• eval_HONENUMBER: {'precision': 1.0, 'recall': 0.8888888888888888, 'f1': 0.9411764705882353, 'number': 9}
• eval_ISCELLANEOUS: {'precision': 0.5521472392638037, 'recall': 0.5660377358490566, 'f1': 0.5590062111801242, 'number': 159}
• eval_MAIL: {'precision': 1.0, 'recall': 0.9803921568627451, 'f1': 0.99009900990099, 'number': 51}
• eval_OCATION: {'precision': 0.8478731074260994, 'recall': 0.9039200614911607, 'f1': 0.875, 'number': 1301}
• eval_P: {'precision': 0.9090909090909091, 'recall': 0.9090909090909091, 'f1': 0.9090909090909091, 'number': 11}
• eval_RL: {'precision': 0.5789473684210527, 'recall': 0.7333333333333333, 'f1': 0.6470588235294117, 'number': 15}
• eval_RODUCT: {'precision': 0.7018739352640545, 'recall': 0.6592, 'f1': 0.6798679867986799, 'number': 625}
• eval_overall_precision: 0.8469
• eval_overall_recall: 0.8683
• eval_overall_f1: 0.8575
• eval_overall_accuracy: 0.9793
• eval_runtime: 38.9411
• eval_samples_per_second: 64.919
• eval_steps_per_second: 16.23
• epoch: 7.0
• step: 22841

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: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1