Update README.md
Browse files
README.md
CHANGED
@@ -3,7 +3,9 @@ language:
|
|
3 |
- en
|
4 |
tags:
|
5 |
- chemistry
|
6 |
-
|
|
|
|
|
7 |
features:
|
8 |
- name: sent1
|
9 |
dtype: string
|
@@ -22,8 +24,14 @@ configs:
|
|
22 |
data_files:
|
23 |
- split: test
|
24 |
path: data/test-*
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
---
|
26 |
|
27 |
# PubChemWikiParagraphsPC
|
28 |
|
29 |
-
This
|
|
|
3 |
- en
|
4 |
tags:
|
5 |
- chemistry
|
6 |
+
- pubchem
|
7 |
+
- wikipedia
|
8 |
+
dataset_info:
|
9 |
features:
|
10 |
- name: sent1
|
11 |
dtype: string
|
|
|
24 |
data_files:
|
25 |
- split: test
|
26 |
path: data/test-*
|
27 |
+
license: cc-by-nc-sa-4.0
|
28 |
+
task_categories:
|
29 |
+
- text-classification
|
30 |
+
pretty_name: PubChem & Wikipedia Paragraphs Pair Classification
|
31 |
+
size_categories:
|
32 |
+
- 1K<n<10K
|
33 |
---
|
34 |
|
35 |
# PubChemWikiParagraphsPC
|
36 |
|
37 |
+
This dataset is for Pair Classification and has three columns: `sent1`, `sent2`, and a binary column `labels`. The `labels` is 1 when `sent1` and `sent2` describe the same entity and 0 when they describe two different entities. The dataset consists of 2048 pairs sourced from [PubChem](https://pubchem.ncbi.nlm.nih.gov/) ([LOTUS](https://lotus.nprod.net/) and [ChEBI](https://www.ebi.ac.uk/chebi/) sources) and English Wikipedia (article summary) with a 90-10% negative-positive ratio.
|