Datasets:

BASF-AI
/

PubChemAISentenceParaphrasePC

Tasks:

Text Classification

Modalities:

Formats:

Languages:

Size:

Tags:

Libraries:

License:

Dataset card Data Studio Files Files and versions Community

hsila commited on Sep 22, 2024

Commit

0f3a22a

·

verified ·

1 Parent(s): eeaad4b

Update README.md

Files changed (1) hide show

README.md +9 -0

README.md CHANGED Viewed

@@ -24,4 +24,13 @@ configs:
     path: data/test-*
 tags:
 - chemistry
 ---

     path: data/test-*
 tags:
 - chemistry
+- pubchem
+license: cc-by-nc-sa-4.0
+pretty_name: PubChem AI Sentence Paraphrased Pair Classification
+size_categories:
+- 1K<n<10K
 ---
+# PubChemAISentenceParaphrasePC
+This dataset consists of sentence pairs, with labels indicating their relationship: a positive label (1) means that one sentence is an AI-generated paraphrase of the other, while a negative label (0) indicates that the two sentences are distinct and unrelated. The paraphrases are created using the `anthropic.claude-3-5-sonnet-20240620-v1:0` model on AWS Bedrock. The dataset is built from entity descriptions sourced from PubChem ([LOTUS](https://lotus.nprod.net/)).