File size: 1,307 Bytes
7b526cd
 
 
 
 
 
 
 
db01c60
 
5ce68e3
db01c60
8dbe548
1ea7433
db01c60
 
9915cd9
db01c60
 
97e8225
 
 
 
21b39f0
0309dda
97e8225
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
---
viewer: false
license:
- apache-2.0
language:
- en
---

**Model Summary**

In order to be able to reproduce GneissWeb, we provide here GneissWeb.Sci_classifier - a science category fastText classifier. This fastText model is used as part of the ensemble filter in GneissWeb to detect documents with science content.

Please refer to the [GneissWeb](https://huggingface.co/datasets/ibm-granite/GneissWeb) for more details.

     **Developers**: IBM Research

     **Release Date**: Feb 21st, 2025

     **License**: Apache 2.0.


**Training Data**

The model is trained on 800k documents, labeled using the [WatsonNLP hierachical categorization](https://www.ibm.com/docs/en/watsonx/saas?topic=catalog-hierarchical-categorization). Please refer to [fastText text classification tutorial](https://fasttext.cc/docs/en/python-module.html) for details. 
Training data is selected as follows:

- *Positive documents*: 400k documents randomly sampled from the documents labeled with science category with a confidence score 0.95 and above.
- *Negative documents*: 400k documents randomly sampled from the documents labeled with any category other than science, education, medical, and technology categories with a confidence score of 0.95 and above.