# stanford-oval/WikiChat

WikiChat is an improved RAG. It stops the hallucination of large language models by retrieving data from a corpus.

Repository: https://github.com/stanford-oval/WikiChat
Canonical: https://ross.abutalabs.com/products/wikichat
Homepage: https://wikichat.genie.stanford.edu
Language: Python
License: Apache-2.0
License Family: permissive
Topics: natural-language-processing, chatbot, nlp, factuality, emnlp2023, rag, llm
Last push: 2026-01-31T04:13:47+00:00

## Health v2 (maintenance only)
Score: 47/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 65, release rhythm 8, longevity 74
- inputs: {"age_days": 1049, "days_push": 214, "days_rel": 491, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1614, forks 146 (observed 2026-08-28T04:05:11.438118+00:00)

## What it is
WikiChat is a retrieval-augmented generation (RAG) framework that grounds LLM chatbot responses in a corpus (Wikipedia by default) to reduce hallucination. It supports custom document indexing with Qdrant, a free Wikipedia search API, distilled lower-latency models, and multi-user deployment via Chainlit.

## Use cases
- build a chatbot that answers factually from wikipedia
- reduce llm hallucination with retrieval-augmented generation
- index my own documents and chat over them
- create a grounded question answering bot over a custom corpus
- distill a smaller model for cheaper factual chatbot responses
- simulate conversations to evaluate chatbot factuality

## When to choose
- you need a chatbot whose answers are grounded in Wikipedia or your own documents
- you want a research-backed RAG pipeline focused on factuality
- you want to build a custom vector index with Qdrant and query it with an LLM

## When to avoid
- you need a general-purpose agent framework with tool use beyond retrieval
- you want a fully hosted turnkey product rather than a self-hosted Python system
- your use case requires real-time data beyond your indexed corpus

## Facets
- artifact type: library
- maturity: active
- function: rag, chatbot, nlp, search-engine, llm-inference
- domain: large-language-models, chatbots
- platform: python, self-hosted
- tags: hallucination-reduction, wikipedia, factuality, qdrant, retrieval-grounding, retrieval-augmented-generation, natural-language-processing, docker

## Member repositories
- stanford-oval/WikiChat (main) score 47

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:11.438118+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 2026-08-30T03:50:31.611935+00:00, confidence not recorded.
  - readme: https://github.com/stanford-oval/WikiChat (fetched 2026-08-28T04:05:11.438118+00:00, sha e750ab5b084d)
  - homepage: https://wikichat.genie.stanford.edu (fetched 2026-08-29T11:22:44.141292+00:00, sha 7c503cee6b1b)
- Data as of 2026-08-30T08:39:29.467469+00:00.