JNLPBA_SciBERT_NER

This model is a fine-tuned version of allenai/scibert_scivocab_uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1472

• Seqeval classification report: precision recall f1-score support

     DNA       0.83      0.89      0.86      2106
     RNA       0.88      0.89      0.88      3516
  

  cell_line 0.74 0.80 0.77 526 cell_type 0.78 0.83 0.80 1475 protein 0.98 0.97 0.98 37428

  micro avg 0.96 0.96 0.96 45051 macro avg 0.84 0.87 0.86 45051

weighted avg 0.96 0.96 0.96 45051

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: 16
• 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
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Seqeval classification report
0.233	1.0	582	0.1513	precision recall f1-score support

     DNA       0.82      0.89      0.85      2106
     RNA       0.87      0.89      0.88      3516

cell_line 0.72 0.79 0.76 526 cell_type 0.79 0.78 0.79 1475 protein 0.98 0.97 0.98 37428

micro avg 0.95 0.95 0.95 45051 macro avg 0.84 0.87 0.85 45051 weighted avg 0.95 0.95 0.95 45051 | | 0.138 | 2.0 | 1164 | 0.1486 | precision recall f1-score support

     DNA       0.85      0.85      0.85      2106
     RNA       0.89      0.87      0.88      3516

cell_line 0.71 0.80 0.75 526 cell_type 0.77 0.82 0.79 1475 protein 0.98 0.97 0.98 37428

micro avg 0.96 0.95 0.95 45051 macro avg 0.84 0.86 0.85 45051 weighted avg 0.96 0.95 0.96 45051 | | 0.1191 | 3.0 | 1746 | 0.1472 | precision recall f1-score support

     DNA       0.83      0.89      0.86      2106
     RNA       0.88      0.89      0.88      3516

cell_line 0.74 0.80 0.77 526 cell_type 0.78 0.83 0.80 1475 protein 0.98 0.97 0.98 37428

micro avg 0.96 0.96 0.96 45051 macro avg 0.84 0.87 0.86 45051 weighted avg 0.96 0.96 0.96 45051 |

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0