|
--- |
|
license: mit |
|
tags: |
|
- fill-mask |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: deberta-v3-large-dapt-scientific-papers-pubmed |
|
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. --> |
|
|
|
# deberta-v3-large-dapt-scientific-papers-pubmed |
|
|
|
This model is a fine-tuned version of [microsoft/deberta-v3-large](https://huggingface.co/microsoft/deberta-v3-large) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 4.4729 |
|
- Accuracy: 0.3510 |
|
|
|
## 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: 1e-06 |
|
- train_batch_size: 12 |
|
- eval_batch_size: 12 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 10000 |
|
- training_steps: 21600 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | |
|
|:-------------:|:-----:|:-----:|:---------------:|:--------:| |
|
| 12.0315 | 0.02 | 500 | 11.6840 | 0.0 | |
|
| 11.0675 | 0.05 | 1000 | 8.9471 | 0.0226 | |
|
| 8.6646 | 0.07 | 1500 | 8.0093 | 0.0344 | |
|
| 8.3625 | 0.09 | 2000 | 7.9624 | 0.0274 | |
|
| 8.2467 | 0.12 | 2500 | 7.6599 | 0.0376 | |
|
| 7.9714 | 0.14 | 3000 | 7.6716 | 0.0316 | |
|
| 7.9852 | 0.16 | 3500 | 7.4535 | 0.0385 | |
|
| 7.7502 | 0.19 | 4000 | 7.4293 | 0.0429 | |
|
| 7.7016 | 0.21 | 4500 | 7.3576 | 0.0397 | |
|
| 7.5789 | 0.23 | 5000 | 7.3124 | 0.0513 | |
|
| 7.4141 | 0.25 | 5500 | 7.1353 | 0.0634 | |
|
| 7.2365 | 0.28 | 6000 | 6.8600 | 0.0959 | |
|
| 7.0725 | 0.3 | 6500 | 6.5743 | 0.1150 | |
|
| 6.934 | 0.32 | 7000 | 6.3674 | 0.1415 | |
|
| 6.7219 | 0.35 | 7500 | 6.3467 | 0.1581 | |
|
| 6.5039 | 0.37 | 8000 | 6.1312 | 0.1815 | |
|
| 6.3096 | 0.39 | 8500 | 5.9080 | 0.2134 | |
|
| 6.1835 | 0.42 | 9000 | 5.8414 | 0.2137 | |
|
| 6.0939 | 0.44 | 9500 | 5.5137 | 0.2553 | |
|
| 6.0457 | 0.46 | 10000 | 5.5881 | 0.2545 | |
|
| 5.8851 | 0.49 | 10500 | 5.5134 | 0.2497 | |
|
| 5.7277 | 0.51 | 11000 | 5.3023 | 0.2699 | |
|
| 5.6183 | 0.53 | 11500 | 5.0074 | 0.3019 | |
|
| 5.4978 | 0.56 | 12000 | 5.1822 | 0.2814 | |
|
| 5.5916 | 0.58 | 12500 | 5.1211 | 0.2808 | |
|
| 5.4749 | 0.6 | 13000 | 4.9126 | 0.2972 | |
|
| 5.3765 | 0.62 | 13500 | 5.0468 | 0.2899 | |
|
| 5.3529 | 0.65 | 14000 | 4.8160 | 0.3037 | |
|
| 5.2993 | 0.67 | 14500 | 4.8598 | 0.3141 | |
|
| 5.2929 | 0.69 | 15000 | 4.9669 | 0.3052 | |
|
| 5.2649 | 0.72 | 15500 | 4.7849 | 0.3270 | |
|
| 5.162 | 0.74 | 16000 | 4.6819 | 0.3357 | |
|
| 5.1639 | 0.76 | 16500 | 4.6056 | 0.3275 | |
|
| 5.1245 | 0.79 | 17000 | 4.5473 | 0.3311 | |
|
| 5.1596 | 0.81 | 17500 | 4.7008 | 0.3212 | |
|
| 5.1346 | 0.83 | 18000 | 4.7932 | 0.3192 | |
|
| 5.1174 | 0.86 | 18500 | 4.7624 | 0.3208 | |
|
| 5.1152 | 0.88 | 19000 | 4.6388 | 0.3274 | |
|
| 5.0852 | 0.9 | 19500 | 4.5247 | 0.3305 | |
|
| 5.0564 | 0.93 | 20000 | 4.6982 | 0.3161 | |
|
| 5.0179 | 0.95 | 20500 | 4.5363 | 0.3389 | |
|
| 5.07 | 0.97 | 21000 | 4.6647 | 0.3307 | |
|
| 5.0781 | 1.0 | 21500 | 4.4729 | 0.3510 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.18.0 |
|
- Pytorch 1.11.0 |
|
- Datasets 2.1.0 |
|
- Tokenizers 0.12.1 |
|
|