Update README.md
Browse files
README.md
CHANGED
@@ -24,21 +24,21 @@ configs:
|
|
24 |
path: data/train-*
|
25 |
---
|
26 |
|
27 |
-
#
|
28 |
|
29 |
## What is this?
|
30 |
|
31 |
-
The following is a high-quality dataset distilled from the Spanish subsection of [RedPajama-Data-v2](https://github.com/togethercomputer/RedPajama-Data), created using the methodology proposed
|
32 |
|
33 |
## Dataset creation
|
34 |
|
35 |
-
In a nutshell, we use Llama-3.1-70B to grade the educational quality of
|
36 |
|
37 |
Here is an overview of the architecture:
|
38 |
|
39 |
![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b15c3f20037ec5d7c91aa6/H5xPOHy_4RhMEDtGvsnTE.png)
|
40 |
|
41 |
-
For more detailed information on how this dataset was created, refer to [our implementation](https://github.com/latam-gpt/llm-data-eval).
|
42 |
|
43 |
## License
|
44 |
|
|
|
24 |
path: data/train-*
|
25 |
---
|
26 |
|
27 |
+
# RedPajama's High Quality Spanish subset
|
28 |
|
29 |
## What is this?
|
30 |
|
31 |
+
The following is a high-quality dataset distilled from the Spanish subsection of [RedPajama-Data-v2](https://github.com/togethercomputer/RedPajama-Data), created using the methodology proposed in [FineWEB-Edu](https://arxiv.org/abs/2406.17557).
|
32 |
|
33 |
## Dataset creation
|
34 |
|
35 |
+
In a nutshell, we use Llama-3.1-70B to grade the educational quality of 550k samples from the original dataset. Then, we used these samples to train a encoder-based classifier, so that it learns to assign a score from 0 to 5. Since this model is cheaper to use than an GPT, we can run it at scale over the entire dataset, thus allowing us to filter a high-quality section from it.
|
36 |
|
37 |
Here is an overview of the architecture:
|
38 |
|
39 |
![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b15c3f20037ec5d7c91aa6/H5xPOHy_4RhMEDtGvsnTE.png)
|
40 |
|
41 |
+
For more detailed information on how this dataset was created, refer to [our open implementation](https://github.com/latam-gpt/llm-data-eval).
|
42 |
|
43 |
## License
|
44 |
|