Vineetiitg
feat(backend): integrate Redis workers, persistent model caching, Cohere V2 fallback, and 5x TTFT fast-path RAG
d816f3a | PROJECT_NAME="Support Docs Copilot" | |
| OPENROUTER_API_KEY=your_openrouter_api_key_here | |
| OPENROUTER_BASE_URL=https://openrouter.ai/api/v1 | |
| LLM_MODEL=deepseek/deepseek-v4-flash | |
| FAST_LLM_MODEL=google/gemini-2.0-flash-lite-preview-02-05:free | |
| FAST_LLM_API_KEY= | |
| FAST_LLM_BASE_URL= | |
| SLOW_LLM_MODEL=deepseek/deepseek-r1 | |
| QDRANT_URL= | |
| QDRANT_LOCATION=./qdrant_data | |
| COLLECTION_NAME=support_docs | |
| DATA_DIR=data/docs | |
| COHERE_API_KEY=your_cohere_api_key_here | |
| RERANKER_PROVIDER=auto | |
| FLASHRANK_MODEL=ms-marco-TinyBERT-L-2-v2 | |
| RERANKER_MODEL=rerank-english-v3.0 | |
| RETRIEVAL_MODE=dense | |
| RETRIEVAL_TOP_K=5 | |
| RERANKER_TOP_N=3 | |
| RERANKER_ENABLED=true | |
| CHUNK_SIZE=500 | |
| CHUNK_OVERLAP=50 | |
| MIN_RELEVANCE_SCORE=0.2 | |
| MAX_CONTEXT_CHARS=12000 | |
| ENABLE_GUARDRAILS=true | |
| ENABLE_RAG_EVAL=false | |
| MAX_QUERY_LENGTH=2000 | |
| RATE_LIMIT_PER_MINUTE=30 | |
| AUTH_ENABLED=false | |
| ADMIN_API_KEY=change-me-admin | |
| USER_API_KEY=change-me-user | |
| # LangSmith Tracing (Optional) | |
| LANGCHAIN_TRACING_V2=false | |
| LANGCHAIN_ENDPOINT=https://api.smith.langchain.com | |
| LANGCHAIN_API_KEY= | |
| LANGCHAIN_PROJECT="Support Docs Copilot" | |