# AkariAsai/OpenScholar

This repository includes the official implementation of OpenScholar: Synthesizing Scientific Literature with Retrieval-augmented LMs.

Repository: https://github.com/AkariAsai/OpenScholar
Canonical: https://ross.abutalabs.com/products/openscholar
Homepage: https://open-scholar.allen.ai/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-08-13T04:16:46+00:00

## Health v2 (maintenance only)
Score: 38/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 36, release rhythm 35, longevity 46
- inputs: {"age_days": 656, "days_push": 385, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1584, forks 167 (observed 2026-08-28T04:05:07.223266+00:00)

## What it is
OpenScholar is a retrieval-augmented language model system from the Allen Institute for AI that answers scientific questions by searching the literature and generating responses grounded in cited papers. The repository provides inference code, training code for a Llama 3.1 8B model, and retriever tooling for offline and online retrieval over scientific corpora.

## Use cases
- answer scientific questions with citations to real papers
- synthesize findings across millions of scientific articles
- build a retrieval-augmented generation pipeline over a research corpus
- host a retrieval server for online literature search
- fine-tune Llama 3.1 8B on scientific QA data
- literature review assistant for researchers

## When to choose
- you need citation-grounded answers to scientific queries rather than generic LLM output
- you want to run or study a full open RAG stack including retriever, inference, and training code
- you are building research tools that must stay current with fast-moving scientific literature

## When to avoid
- you need a polished end-user product rather than research code you assemble yourself
- you lack GPU resources or API keys for the underlying language models and retrieval infrastructure
- your domain is not scientific literature and you need a general-purpose RAG framework

## Facets
- artifact type: application
- maturity: active
- function: rag, llm-inference, llm-training, search-engine, nlp
- domain: large-language-models, artificial-intelligence, data-science
- platform: python, cli
- tags: scientific-literature, retrieval-augmented-generation, llama-3.1, citation-grounded-answers, research-assistant, allenai, scholarqa, natural-language-processing, search, web-server, gpu, docker

## Member repositories
- AkariAsai/OpenScholar (main) score 38

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:07.223266+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:55:40.046361+00:00, confidence not recorded.
  - readme: https://github.com/AkariAsai/OpenScholar (fetched 2026-08-28T04:05:07.223266+00:00, sha 6a704fd81c69)
  - homepage: https://open-scholar.allen.ai/ (fetched 2026-08-29T11:26:25.564869+00:00, sha 1f23b7c2fe45)
- Data as of 2026-08-30T08:39:29.467469+00:00.