Hugging Face's logo

Datasets:
SciPhi
/
AgentSearch-V1

Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10K<n<100K
ArXiv:
Tags:
Libraries:
Datasets
Dask
Croissant
Dataset card Viewer Files Community
3
AgentSearch-V1 / README.md
lhoestq's picture
lhoestq HF staff
Set data_files to `**/*.parquet` by default
fc7439e
preview code
|
raw history blame
No virus
4.85 kB
metadata
language:
  - en
size_categories:
  - 1B<n<10B
task_categories:
  - text-generation
pretty_name: AgentSearch-V1
configs:
  - config_name: default
    data_files:
      - split: train
        path: '**/*.parquet'

Important Notice

This dataset is just a sample. The real dataset will be uploaded after New Year's 2024. This early release is intended for Agent Search launching today, but the data is not yet finalized.

Getting Started

The AgentSearch-V1 dataset includes over one billion embeddings sourced from over 50 million high-quality documents. This extensive collection encompasses the majority of content from sources like Arxiv, Wikipedia, Project Gutenberg, and includes quality-filtered CC data.

To access and utilize the AgentSearch-V1 dataset, you can stream it via HuggingFace with the following Python code:

from datasets import load_dataset
# To stream the entire dataset:
ds = load_dataset("SciPhi/AgentSearch-V1", data_files="**/*", streaming=True)

# Optional, stream just the "arxiv" dataset
ds = load_dataset("SciPhi/AgentSearch-V1", data_files="arxiv/*", streaming=True)

A full set of scripts to recreate the dataset from scratch can be found here. Synthesizer offers direct integration with AgentSearch and top LLM providers.

Dataset Summary

We take a similar approach to RedPajama-v1 and divide AgentSearch into a number of categories.

Dataset Token Count
Books TBD
ArXiv TBD
Wikipedia TBD
StackExchange TBD
OpenMath TBD
Filtered Crawl TBD
Total TBD

Languages

English.

Dataset Structure

The raw dataset structure is as follows:

{
    "url": ...,
    "title": ...,
    "metadata": {"url": "...", "timestamp": "...", "source": "...", "language": "...", ...},
    "text_chunks": ...,
    "embeddings": ...,
    "dataset": "github" | "books" | "arxiv" | "wikipedia" | "stackexchange" | "open-math" | "filtered-rp2"
}

The indexed dataset can be downloaded directly and is structured as a qdrant database dump, each entry has meta data {"url", "vector"}. In addition, there is a corresponding sqlite dataset which contains the mapping from urls onto embeddings, text chunks, and other metadata.

Dataset Creation

This dataset was created as a step towards making humanities most important knowledge locally searchable and LLM optimal. It was created by filtering, cleaning, and augmenting locally publicly available datasets.

To cite our work, please use the following:

@software{SciPhi2023AgentSearch,
  author = {SciPhi},
  title = {AgentSearch [ΨΦ]: A Comprehensive Agent-First Framework and Dataset for Webscale Search},
  year = {2023},
  url = {https://github.com/SciPhi-AI/agent-search}
}

Source Data 

@ONLINE{wikidump,
    author = "Wikimedia Foundation",
    title  = "Wikimedia Downloads",
    url    = "https://dumps.wikimedia.org"
}

@misc{paster2023openwebmath,
      title={OpenWebMath: An Open Dataset of High-Quality Mathematical Web Text},
      author={Keiran Paster and Marco Dos Santos and Zhangir Azerbayev and Jimmy Ba},
      year={2023},
      eprint={2310.06786},
      archivePrefix={arXiv},
      primaryClass={cs.AI}
}

@software{together2023redpajama,
  author = {Together Computer},
  title = {RedPajama: An Open Source Recipe to Reproduce LLaMA training dataset},
  month = April,
  year = 2023,
  url = {https://github.com/togethercomputer/RedPajama-Data}
}

License

Please refer to the licenses of the data subsets you use.