BigHugger
GH Repository · Marker-Inc-Korea

AutoRAG

AutoRAG: Now your agent can find anything in your computer. It gets smarter if you are using it frequently.

stars
5,073
30-day movement
+72/day
Related entries
62
Connections
2
typescriptnodemakeevalrag-evaluationragdocument-parsernode/bunautomlTypeScriptqabenchmarkingretrieval-augmented-generationanalysisembeddingsllm-evaluationevaluationllm-opspipelineopen-sourcepythonllmopsoptimization

AutoRAG is an open-source RAG evaluation and optimization tool from Marker-Inc-Korea. Its topics cover AutoML-style pipeline optimization, benchmarking, document parsing, embeddings, and LLM evaluation, with the tagline that an agent can find anything on your computer and improves with frequent use.

You want to evaluate and tune RAG pipelines rather than guess at retrieval and generation quality.

Use it to

  • Benchmark RAG pipeline configurations
  • Optimize retrieval and embedding choices automatically
  • Parse documents for QA datasets
  • Evaluate LLM answers against ground truth
  • Track RAG ops metrics over time

For Teams building and evaluating RAG systems

Role
eval
Language
TypeScript
Licence
custom (licence file present)
Forks
435
Open issues
127
Last push
2026-09-17
Latest release
v0.0.2 · 2024-02-03
Skills shipped
2
topicsragevaluationbenchmarkingoptimizationllmretrieval