|
--- |
|
library_name: transformers |
|
license: apache-2.0 |
|
base_model: distilbert-base-uncased-distilled-squad |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- f1 |
|
model-index: |
|
- name: distilbert-qasports |
|
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. --> |
|
|
|
# distilbert-qasports |
|
|
|
This model is a fine-tuned version of [distilbert-base-uncased-distilled-squad](https://huggingface.co/distilbert-base-uncased-distilled-squad) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.4015 |
|
- Exact: 76.8499 |
|
- F1: 81.2744 |
|
- Total: 15041 |
|
- Hasans Exact: 76.8499 |
|
- Hasans F1: 81.2744 |
|
- Hasans Total: 15041 |
|
- Best Exact: 76.8499 |
|
- Best Exact Thresh: 0.0 |
|
- Best F1: 81.2744 |
|
- Best F1 Thresh: 0.0 |
|
|
|
## 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-05 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 16 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 32 |
|
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
|
- lr_scheduler_type: linear |
|
- num_epochs: 50 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Exact | F1 | Total | Hasans Exact | Hasans F1 | Hasans Total | Best Exact | Best Exact Thresh | Best F1 | Best F1 Thresh | |
|
|:-------------:|:------:|:----:|:---------------:|:-------:|:-------:|:-----:|:------------:|:---------:|:------------:|:----------:|:-----------------:|:-------:|:--------------:| |
|
| 0.6883 | 0.1325 | 500 | 0.6008 | 74.7690 | 79.7676 | 15041 | 74.7690 | 79.7676 | 15041 | 74.7690 | 0.0 | 79.7676 | 0.0 | |
|
| 0.5738 | 0.2649 | 1000 | 0.5474 | 75.3407 | 80.2816 | 15041 | 75.3407 | 80.2816 | 15041 | 75.3407 | 0.0 | 80.2816 | 0.0 | |
|
| 0.5853 | 0.3974 | 1500 | 0.5259 | 75.3873 | 80.2217 | 15041 | 75.3873 | 80.2217 | 15041 | 75.3873 | 0.0 | 80.2217 | 0.0 | |
|
| 0.588 | 0.5298 | 2000 | 0.4904 | 76.3978 | 81.0881 | 15041 | 76.3978 | 81.0881 | 15041 | 76.3978 | 0.0 | 81.0881 | 0.0 | |
|
| 0.5214 | 0.6623 | 2500 | 0.4764 | 76.8366 | 81.4327 | 15041 | 76.8366 | 81.4327 | 15041 | 76.8366 | 0.0 | 81.4327 | 0.0 | |
|
| 0.4813 | 0.7947 | 3000 | 0.4586 | 76.9763 | 81.6042 | 15041 | 76.9763 | 81.6042 | 15041 | 76.9763 | 0.0 | 81.6042 | 0.0 | |
|
| 0.5032 | 0.9272 | 3500 | 0.4323 | 76.7835 | 81.4041 | 15041 | 76.7835 | 81.4041 | 15041 | 76.7835 | 0.0 | 81.4041 | 0.0 | |
|
| 0.3549 | 1.0596 | 4000 | 0.4349 | 76.8632 | 81.2899 | 15041 | 76.8632 | 81.2899 | 15041 | 76.8632 | 0.0 | 81.2899 | 0.0 | |
|
| 0.4053 | 1.1921 | 4500 | 0.4199 | 76.9630 | 81.3741 | 15041 | 76.9630 | 81.3741 | 15041 | 76.9630 | 0.0 | 81.3741 | 0.0 | |
|
| 0.3549 | 1.3245 | 5000 | 0.4372 | 77.0427 | 81.6167 | 15041 | 77.0427 | 81.6167 | 15041 | 77.0427 | 0.0 | 81.6167 | 0.0 | |
|
| 0.3707 | 1.4570 | 5500 | 0.4254 | 77.0560 | 81.5058 | 15041 | 77.0560 | 81.5058 | 15041 | 77.0560 | 0.0 | 81.5058 | 0.0 | |
|
| 0.3728 | 1.5894 | 6000 | 0.4086 | 76.9031 | 81.4012 | 15041 | 76.9031 | 81.4012 | 15041 | 76.9031 | 0.0 | 81.4012 | 0.0 | |
|
| 0.4117 | 1.7219 | 6500 | 0.4029 | 76.8233 | 81.4108 | 15041 | 76.8233 | 81.4108 | 15041 | 76.8233 | 0.0 | 81.4108 | 0.0 | |
|
| 0.3785 | 1.8543 | 7000 | 0.3979 | 77.0427 | 81.4664 | 15041 | 77.0427 | 81.4664 | 15041 | 77.0427 | 0.0 | 81.4664 | 0.0 | |
|
| 0.3564 | 1.9868 | 7500 | 0.4015 | 76.8499 | 81.2744 | 15041 | 76.8499 | 81.2744 | 15041 | 76.8499 | 0.0 | 81.2744 | 0.0 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.48.2 |
|
- Pytorch 2.0.0+cu117 |
|
- Datasets 3.2.0 |
|
- Tokenizers 0.21.0 |
|
|
