thiagolaitz
commited on
Commit
•
e89e1f2
1
Parent(s):
ee2d86d
Update README.md
Browse files
README.md
CHANGED
@@ -6,4 +6,22 @@ language:
|
|
6 |
metrics:
|
7 |
- perplexity
|
8 |
library_name: transformers
|
9 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
metrics:
|
7 |
- perplexity
|
8 |
library_name: transformers
|
9 |
+
---
|
10 |
+
|
11 |
+
This model is a Portuguese fine-tuned version of the [facebook/opt-125m](https://huggingface.co/facebook/opt-125m). It has undergone additional causal language modeling pre-training with a context size of 512, using an extra 300 million tokens in Portuguese (sampled from mc4). The Wandb report is publicly available at [here](https://api.wandb.ai/links/thiagolaitz1/ths2zi4c). The code for training using Colab pro (A100 - 40GB) can be found [here](https://github.com/thiagolaitz/IA368-search-engines/blob/main/Project%2004/opt_125m_pt_finetuning.ipynb). The total cost for training this model was R$17.40 or $3.37 USD (as of March 2023).
|
12 |
+
|
13 |
+
Deterministic use:
|
14 |
+
```python
|
15 |
+
from transformers import pipeline
|
16 |
+
|
17 |
+
generator = pipeline('text-generation', model="thiagolaitz/opt-125m-pt-finetuned", max_length=30)
|
18 |
+
generator("Eles brincaram o dia inteiro sob o sol quente, mas")
|
19 |
+
```
|
20 |
+
|
21 |
+
Top-k sampling:
|
22 |
+
```python
|
23 |
+
from transformers import pipeline
|
24 |
+
|
25 |
+
generator = pipeline('text-generation', model="thiagolaitz/opt-125m-pt-finetuned", do_sample=True, max_length=30)
|
26 |
+
generator("Eles brincaram o dia inteiro sob o sol quente, mas")
|
27 |
+
```
|