metadata
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: bart-mlm-pubmed
results: []
bart-mlm-pubmed
This model is a fine-tuned version of facebook/bart-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7223
- Rouge2 Precision: 0.6577
- Rouge2 Recall: 0.5167
- Rouge2 Fmeasure: 0.5667
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
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge2 Precision | Rouge2 Recall | Rouge2 Fmeasure |
|---|---|---|---|---|---|---|
| 1.0322 | 1.0 | 663 | 0.7891 | 0.6398 | 0.4995 | 0.5499 |
| 0.8545 | 2.0 | 1326 | 0.7433 | 0.6464 | 0.5061 | 0.5559 |
| 0.758 | 3.0 | 1989 | 0.7299 | 0.6479 | 0.5041 | 0.5553 |
| 0.6431 | 4.0 | 2652 | 0.7185 | 0.656 | 0.5106 | 0.562 |
| 0.6058 | 5.0 | 3315 | 0.7126 | 0.6542 | 0.5153 | 0.5648 |
| 0.5726 | 6.0 | 3978 | 0.7117 | 0.6572 | 0.5176 | 0.5674 |
| 0.5168 | 7.0 | 4641 | 0.7150 | 0.659 | 0.5161 | 0.5668 |
| 0.5011 | 8.0 | 5304 | 0.7220 | 0.6574 | 0.517 | 0.5669 |
| 0.4803 | 9.0 | 5967 | 0.7208 | 0.6579 | 0.5167 | 0.5668 |
| 0.4577 | 10.0 | 6630 | 0.7223 | 0.6577 | 0.5167 | 0.5667 |
Framework versions
- Transformers 4.12.3
- Pytorch 1.9.0+cu111
- Datasets 1.15.1
- Tokenizers 0.10.3