mazzaqq's picture
Update README.md
55db905
metadata
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: 8Agos
    results: []

roberta_finetuned_astronomicalNER

This model is a fine-tuned version of xlm-roberta-large-finetuned-conll03-english for NER on astronomical objects. The dataset comes from the Shared Task DEAL: Detecting Entities in the Astrophysics Literature

The model achieves the following results on the evaluation set:

  • Loss: 0.1416
  • Precision: 0.7659
  • Recall: 0.7986
  • F1: 0.7819
  • Accuracy: 0.9640

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 2e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 1.0 176 0.1571 0.7362 0.7788 0.7569 0.9593
No log 2.0 352 0.1416 0.7529 0.7831 0.7677 0.9624
0.1109 3.0 528 0.1416 0.7659 0.7986 0.7819 0.9640

Framework versions

  • Transformers 4.21.1
  • Pytorch 1.12.0+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1