{"id":30780344,"url":"https://github.com/codewiththomas/researchrag","last_synced_at":"2026-03-07T03:02:27.152Z","repository":{"id":283963376,"uuid":"953413610","full_name":"codewiththomas/ResearchRAG","owner":"codewiththomas","description":"Ein modulares RAG-System mit autoamtischer Evaluationsfunktion zum Testen verschiedener RAG-Konfigurationen.","archived":false,"fork":false,"pushed_at":"2025-08-29T08:08:20.000Z","size":111654,"stargazers_count":1,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-09-17T09:56:16.405Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/codewiththomas.png","metadata":{"files":{"readme":"Readme.MD","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-03-23T10:05:43.000Z","updated_at":"2025-08-29T08:07:09.000Z","dependencies_parsed_at":null,"dependency_job_id":"87fd13b5-870d-48fc-82e7-17c9c0ddc5f0","html_url":"https://github.com/codewiththomas/ResearchRAG","commit_stats":null,"previous_names":["codewiththomas/fom.bigdataanalyseprojekt","codewiththomas/researchrag"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/codewiththomas/ResearchRAG","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/codewiththomas%2FResearchRAG","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/codewiththomas%2FResearchRAG/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/codewiththomas%2FResearchRAG/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/codewiththomas%2FResearchRAG/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/codewiththomas","download_url":"https://codeload.github.com/codewiththomas/ResearchRAG/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/codewiththomas%2FResearchRAG/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30206339,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-06T19:07:06.838Z","status":"online","status_checked_at":"2026-03-07T02:00:06.765Z","response_time":53,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-09-05T07:10:33.835Z","updated_at":"2026-03-07T03:02:27.130Z","avatar_url":"https://github.com/codewiththomas.png","language":"Python","readme":"# Research RAG\n\nEin modulates RAG-System, welches eine systematische Evaluation ermöglicht.\n\n## Nutzung\n\n1. Clone\n\n2. Branch\n\n3. Virtual Environment erstellen (.venv)\n\n```bash\npy -3.12 -m venv .venv\n```\n\n4. Virtual Environment aktivieren\n\n```bash\n./.venv/bin/activate\n```\n\n4. Requirements installieren\n\n```bash\npip install -r requirements.txt\n```\n\n5. `.env` erstellen (an `.env.example`orientieren; OPENAI_API_KEY setzen)\n\n6. Ausführen\n\n```bash\n python src/rag/main.py --config configs/000_baseline.yaml --dataset data/output/dsgvo_crawled_2025-08-20_1824.jsonl --num-qa 5\n ```\n\n---\n\n## Module\n\n### Indexer\n\n### Sprachmodell\n\nResearch-RAG ermöglicht die Einbindung verschiedener Sprachmodelle.\n\n```bash\nollama pull mixtral:8x7b\nollama pull mixtral:8x22b\n```\n\n\n## Evaluation\n\n### Ziel: zweistufige Evaluation des RAG-Systems.\n\n1. Retrieval-Ebene (IR-Metriken): Grid-Search über Chunking/Parameter\n\n2. Antwort-Ebene (RAGAS + DSGVO-Score): Vergleich von SLMs\n\n### Voraussetzungen (Baseline)\n\nIn configs/000_baseline.yaml festlegen:\n\n- dataset.path und dataset.evaluation_subset_size (für Smoke-Tests z. B. 1) \n\n- embedding.model_name und embedding.abbr (für Dateinamen) \n\n- LLM-Defaults (z. B. llm.model, llm.temperature, llm.max_tokens) \n\n- Kontextgrenze: pipeline.max_context_length (= max_content_size) \n\n### Retrieval-Grid (IR-Metriken)\n\nVariierte Parameter:\n- chunking.type\n- chunking.chunk_size\n- retrieval.top_k\n- retrieval.similarity_threshold\n- dataset.grouping.enabled.\nErgebnisse (Detail + Summary, mit Zeitstempel) landen in results/runs/. \n\n```bash\n# Beispiel: alle Retrieval-Kombinationen ausführen\npython src/rag/retrieval_grid_search.py\n```\n\nHinweis: Das Skript liest die Baseline, überschreibt nur die o. g. Retrieval-Parameter und speichert pro Run JSON-Dateien (Zeitstempel kommt vom Evaluator). \n\nLLM-Grid (Antwort-Ebene)","funding_links":[],"categories":[],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcodewiththomas%2Fresearchrag","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcodewiththomas%2Fresearchrag","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcodewiththomas%2Fresearchrag/lists"}