Fetching metadata from the HF Docker repository...
-
feat: multi-key rotation for Gemini (3 keys) and Groq (3 keys)
=
- Frontend: GeminiKeyManager with round-robin rotation + withRetry on 429
Reads VITE_GEMINI_API_KEY_1/2/3 (falls back to VITE_GEMINI_API_KEY)
- Backend: add GROQ_API_KEY_3 to Config + GroqKeyManager (3 keys)
- Dockerfile.frontend: build-args for 3 Gemini keys
- rag_project/Dockerfile: ENV for GROQ_API_KEY_3
- render.yaml: GROQ_API_KEY_3 envVar entry
- .env.example files updated (keys 1-3, no secrets committed)
-
feat: full optimization - Groq LLM, disease cache, deploy configs
mnhat19
- Switch backend LLM from Gemini to Groq (llama-3.3-70b-versatile)
- Add GroqKeyManager: round-robin key rotation with 429 cooldown
- Add DiseaseCache: SQLite 7-day TTL cache - 0 LLM calls on cache HIT
- SessionStore: WAL mode + cleanup_expired() on startup
- Fix nhan_xet_tong_quan field in evaluation JSON (was 'nhan_xet')
- Restore proper Vietnamese diacritics in all LLM prompts
- Migrate Tailwind from CDN to PostCSS pipeline
- Add tenacity retry on rate limit errors
- ragService.ts: use VITE_RAG_API_URL env var (not hardcoded localhost)
- Add render.yaml (Render.com free tier backend deploy)
- Add vercel.json (Vercel frontend deploy)
- Add rag_project/Dockerfile (HuggingFace Spaces alternative)
- Update .gitignore: exclude secrets, DB files, training data
-
Push all code to huy_xau_trai repository
huydt11502
-
Add RAG integration: Flask API server, disease selector, evaluation system with improved case generation
huydt11502
