metadata
license: apache-2.0
base_model: facebook/bart-base
tags:
- generated_from_trainer
model-index:
- name: bart-with-pubmed-noise-data-0.1-v2
results: []
bart-with-pubmed-noise-data-0.1-v2
This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2115
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
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.4161 | 0.11 | 500 | 0.3441 |
| 0.342 | 0.21 | 1000 | 0.3091 |
| 0.2694 | 0.32 | 1500 | 0.2969 |
| 0.3792 | 0.43 | 2000 | 0.2712 |
| 0.3219 | 0.54 | 2500 | 0.2601 |
| 0.3001 | 0.64 | 3000 | 0.2574 |
| 0.2606 | 0.75 | 3500 | 0.2489 |
| 0.2716 | 0.86 | 4000 | 0.2415 |
| 0.2714 | 0.96 | 4500 | 0.2382 |
| 0.2072 | 1.07 | 5000 | 0.2429 |
| 0.2111 | 1.18 | 5500 | 0.2377 |
| 0.1977 | 1.28 | 6000 | 0.2455 |
| 0.2171 | 1.39 | 6500 | 0.2309 |
| 0.1853 | 1.5 | 7000 | 0.2314 |
| 0.2436 | 1.61 | 7500 | 0.2269 |
| 0.171 | 1.71 | 8000 | 0.2220 |
| 0.2032 | 1.82 | 8500 | 0.2226 |
| 0.2028 | 1.93 | 9000 | 0.2175 |
| 0.1448 | 2.03 | 9500 | 0.2227 |
| 0.1447 | 2.14 | 10000 | 0.2216 |
| 0.1516 | 2.25 | 10500 | 0.2200 |
| 0.1294 | 2.35 | 11000 | 0.2197 |
| 0.1569 | 2.46 | 11500 | 0.2157 |
| 0.1505 | 2.57 | 12000 | 0.2160 |
| 0.152 | 2.68 | 12500 | 0.2151 |
| 0.1588 | 2.78 | 13000 | 0.2117 |
| 0.1451 | 2.89 | 13500 | 0.2134 |
| 0.1644 | 3.0 | 14000 | 0.2115 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1