|
--- |
|
license: apache-2.0 |
|
base_model: distilgpt2 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: distilgpt2-finetuned-wikitext2 |
|
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. --> |
|
|
|
# distilgpt2-finetuned-wikitext2 |
|
|
|
This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 3.6419 |
|
|
|
## 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: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 3.0 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 3.7712 | 1.0 | 2334 | 3.6691 | |
|
| 3.6526 | 2.0 | 4668 | 3.6486 | |
|
| 3.6075 | 3.0 | 7002 | 3.6419 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.32.0 |
|
- Pytorch 2.0.1+cu118 |
|
- Datasets 2.14.4 |
|
- Tokenizers 0.13.3 |
|
|
