BASF-AI/PubChemAISentenceParaphrasePC

--- language: - en task_categories: - text-classification dataset_info: features: - name: sent1 dtype: string - name: sent2 dtype: string - name: labels dtype: int64 splits: - name: test num_bytes: 381014 num_examples: 2048 download_size: 184976 dataset_size: 381014 configs: - config_name: default data_files: - split: test path: data/test-* tags: - chemistry - pubchem - chemteb license: cc-by-nc-sa-4.0 pretty_name: PubChem AI Paraphrased Sentence Pair Classification size_categories: - 1K<n<10K --- # PubChem AI Paraphrased Sentence Pair Classification 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 ([ChEBI](https://www.ebi.ac.uk/chebi/) source).