File size: 2,070 Bytes
d3bdcd7 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
---
license: mit
tags:
- generated_from_trainer
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: tmvar_5e-05_0404_ES6_strict_tok1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# tmvar_5e-05_0404_ES6_strict_tok1
This model is a fine-tuned version of [microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0372
- Precision: 0.7742
- Recall: 0.8528
- F1: 0.8116
- Accuracy: 0.9906
## 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: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- training_steps: 2000
### Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 0.3642 | 0.49 | 25 | 0.0757 | 0.0 | 0.0 | 0.0 | 0.9727 |
| 0.0672 | 0.98 | 50 | 0.0660 | 0.6397 | 0.4416 | 0.5225 | 0.9841 |
| 0.0347 | 1.47 | 75 | 0.0357 | 0.7129 | 0.7310 | 0.7218 | 0.9888 |
| 0.0292 | 1.96 | 100 | 0.0255 | 0.7630 | 0.8173 | 0.7892 | 0.9903 |
| 0.012 | 2.45 | 125 | 0.0325 | 0.6923 | 0.8223 | 0.7517 | 0.9903 |
| 0.0087 | 2.94 | 150 | 0.0372 | 0.7742 | 0.8528 | 0.8116 | 0.9906 |
### Framework versions
- Transformers 4.27.4
- Pytorch 2.0.0+cu118
- Datasets 2.11.0
- Tokenizers 0.13.3
|