metadata
license: cc0-1.0
arxiv-abstracts-2021-embeddings-10000
This repository contains a subset of the gfissore/arxiv-abstracts-2021 dataset, specifically the first 10,000 samples. It includes embeddings generated from three different models.
Dataset Structure
Each dataset contains the following columns:
- id: Unique identifier for each entry.
- content: The abstract text from the arXiv paper.
- categories: The categories associated with the paper.
- embedding: The embedding representation of the abstract.
Available Datasets
The repository includes three datasets, each corresponding to a different embedding model:
data/arxiv-abstract-arcticlarge.parquet: Embeddings generated using the Arctic Large model. Model card here: Snowflake/snowflake-arctic-embed-m-v2.0data/arxiv-abstract-arcticmedium.parquet: Embeddings generated using the Arctic Medium model. Model card here: Snowflake/snowflake-arctic-embed-m-v2.0data/arxiv-abstract-minilm.parquet: Embeddings generated using the MiniLM model. Model card here: sentence-transformers/all-MiniLM-L6-v2