license: mit
base_model: microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext
tags:
- generated_from_trainer
model-index:
- name: BioNLP13CG_PubMedBERT_NER
results: []
BioNLP13CG_PubMedBERT_NER
This model is a fine-tuned version of microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext on the None dataset. It achieves the following results on the evaluation set:
Loss: 0.2066
Seqeval classification report: precision recall f1-score support
Amino_acid 0.78 0.81 0.79 301 Anatomical_system 0.00 0.00 0.00 3 Cancer 0.00 0.00 0.00 37 Cell 0.79 0.85 0.82 446 Cellular_component 0.00 0.00 0.00 19
Developing_anatomical_structure 0.55 0.78 0.65 399 Gene_or_gene_product 0.68 0.41 0.51 128 Immaterial_anatomical_entity 0.00 0.00 0.00 45 Multi-tissue_structure 0.25 0.02 0.04 98 Organ 0.00 0.00 0.00 19 Organism 0.90 0.93 0.92 1108 Organism_subdivision 0.71 0.12 0.21 120 Organism_substance 0.62 0.59 0.60 128 Pathological_formation 0.00 0.00 0.00 41 Simple_chemical 0.87 0.86 0.86 4397 Tissue 0.90 0.93 0.91 1790
micro avg 0.84 0.83 0.84 9079
macro avg 0.44 0.39 0.39 9079
weighted avg 0.83 0.83 0.82 9079
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 |
|---|---|---|---|---|
| No log | 0.99 | 95 | 0.3390 | precision recall f1-score support |
Amino_acid 0.81 0.10 0.18 301
Anatomical_system 0.00 0.00 0.00 3
Cancer 0.00 0.00 0.00 37
Cell 0.82 0.76 0.79 446
Cellular_component 0.00 0.00 0.00 19
Developing_anatomical_structure 0.90 0.07 0.13 399 Gene_or_gene_product 0.00 0.00 0.00 128 Immaterial_anatomical_entity 0.00 0.00 0.00 45 Multi-tissue_structure 0.00 0.00 0.00 98 Organ 0.00 0.00 0.00 19 Organism 0.64 0.86 0.73 1108 Organism_subdivision 0.00 0.00 0.00 120 Organism_substance 0.00 0.00 0.00 128 Pathological_formation 0.00 0.00 0.00 41 Simple_chemical 0.83 0.79 0.81 4397 Tissue 0.74 0.91 0.82 1790
micro avg 0.77 0.71 0.74 9079
macro avg 0.30 0.22 0.22 9079
weighted avg 0.73 0.71 0.69 9079
| | No log | 2.0 | 191 | 0.2209 | precision recall f1-score support
Amino_acid 0.76 0.75 0.76 301
Anatomical_system 0.00 0.00 0.00 3
Cancer 0.00 0.00 0.00 37
Cell 0.78 0.87 0.82 446
Cellular_component 0.00 0.00 0.00 19
Developing_anatomical_structure 0.52 0.75 0.61 399 Gene_or_gene_product 0.65 0.24 0.35 128 Immaterial_anatomical_entity 0.00 0.00 0.00 45 Multi-tissue_structure 0.00 0.00 0.00 98 Organ 0.00 0.00 0.00 19 Organism 0.89 0.92 0.91 1108 Organism_subdivision 0.50 0.05 0.09 120 Organism_substance 0.61 0.52 0.56 128 Pathological_formation 0.00 0.00 0.00 41 Simple_chemical 0.86 0.86 0.86 4397 Tissue 0.87 0.93 0.90 1790
micro avg 0.83 0.82 0.83 9079
macro avg 0.40 0.37 0.37 9079
weighted avg 0.81 0.82 0.81 9079
| | No log | 2.98 | 285 | 0.2066 | precision recall f1-score support
Amino_acid 0.78 0.81 0.79 301
Anatomical_system 0.00 0.00 0.00 3
Cancer 0.00 0.00 0.00 37
Cell 0.79 0.85 0.82 446
Cellular_component 0.00 0.00 0.00 19
Developing_anatomical_structure 0.55 0.78 0.65 399 Gene_or_gene_product 0.68 0.41 0.51 128 Immaterial_anatomical_entity 0.00 0.00 0.00 45 Multi-tissue_structure 0.25 0.02 0.04 98 Organ 0.00 0.00 0.00 19 Organism 0.90 0.93 0.92 1108 Organism_subdivision 0.71 0.12 0.21 120 Organism_substance 0.62 0.59 0.60 128 Pathological_formation 0.00 0.00 0.00 41 Simple_chemical 0.87 0.86 0.86 4397 Tissue 0.90 0.93 0.91 1790
micro avg 0.84 0.83 0.84 9079
macro avg 0.44 0.39 0.39 9079
weighted avg 0.83 0.83 0.82 9079
|
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0