|
--- |
|
license: apache-2.0 |
|
base_model: bert-base-uncased |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- wikitext |
|
model-index: |
|
- name: run_opt |
|
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. --> |
|
|
|
# run_opt |
|
|
|
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the wikitext dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0165 |
|
|
|
## 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: 0.0003 |
|
- train_batch_size: 64 |
|
- eval_batch_size: 64 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 512 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_steps: 100 |
|
- num_epochs: 10 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 8.562 | 0.55 | 50 | 6.9697 | |
|
| 6.63 | 1.1 | 100 | 6.3436 | |
|
| 5.938 | 1.65 | 150 | 5.1110 | |
|
| 3.0597 | 2.19 | 200 | 1.4150 | |
|
| 0.7989 | 2.74 | 250 | 0.3477 | |
|
| 0.2227 | 3.29 | 300 | 0.1284 | |
|
| 0.0925 | 3.84 | 350 | 0.0640 | |
|
| 0.0475 | 4.39 | 400 | 0.0412 | |
|
| 0.0314 | 4.94 | 450 | 0.0304 | |
|
| 0.0217 | 5.49 | 500 | 0.0246 | |
|
| 0.0181 | 6.04 | 550 | 0.0215 | |
|
| 0.0146 | 6.58 | 600 | 0.0194 | |
|
| 0.0132 | 7.13 | 650 | 0.0182 | |
|
| 0.012 | 7.68 | 700 | 0.0174 | |
|
| 0.0114 | 8.23 | 750 | 0.0169 | |
|
| 0.011 | 8.78 | 800 | 0.0167 | |
|
| 0.0108 | 9.33 | 850 | 0.0166 | |
|
| 0.0106 | 9.88 | 900 | 0.0165 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.33.1 |
|
- Pytorch 1.12.1 |
|
- Datasets 2.14.6 |
|
- Tokenizers 0.13.3 |
|
|
