39 lines
845 B
Markdown
39 lines
845 B
Markdown
|
---
|
||
|
|
widget:
|
||
|
|
- text: Em uma bela manhã de
|
||
|
|
- text: Em uma cidade tão grande como
|
||
|
|
- text: Maria e Joana são
|
||
|
|
license: mit
|
||
|
|
datasets:
|
||
|
|
- mc4
|
||
|
|
language:
|
||
|
|
- pt
|
||
|
|
metrics:
|
||
|
|
- perplexity
|
||
|
|
library_name: transformers
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
---
|
||
|
|
# OPT-125M finetuned Portuguese
|
||
|
|
|
||
|
|
Fine-tuning the [OPT-125M](https://huggingface.co/facebook/opt-125m) model on a reduced corpus of mc4-Portuguese with approximately 300M tokens.
|
||
|
|
|
||
|
|
###### Hyper-parameters
|
||
|
|
|
||
|
|
|
||
|
|
- learning_rate = 5e-5
|
||
|
|
- batch_size = 32
|
||
|
|
- warmup = 500
|
||
|
|
- seq_length = 512
|
||
|
|
- num_train_epochs = 2.0
|
||
|
|
|
||
|
|
With an A100 with 40GB of RAM, the training took around 3 hours
|
||
|
|
|
||
|
|
**Perplexity:** 9.4
|
||
|
|
|
||
|
|
## Sample Use
|
||
|
|
|
||
|
|
```python
|
||
|
|
from transformers import pipeline
|
||
|
|
generator = pipeline('text-generation', model='Mirelle/opt-125M-pt-br-finetuned', max_length=100, do_sample=True)
|
||
|
|
generator("Em uma bela manhã de")
|
||
|
|
```
|