xXArjunXx/MetaHackathon

Updated Files

updated inference.py

updated

updated

updated

Updated

Updated File

Fix: Ensure all grader scores strictly within (0, 1) open interval

updated graded score

Updated

updated

Updated

commit

Fix Phase 2: Guarantee LiteLLM proxy detection

updated inference.py

Fix proxy env handling for local and validator runs

Fix Phase 2 failure: Add MODEL_NAME validation for LLM proxy API calls

updated inference.py

Updated

fix: read env vars directly from os.environ for validator compatibility

fix: prioritize injected validator env over local .env

fix: sync build_client() fix to MetaHackathon/inference.py

git commit -m "fix: add structured [START]/[STEP]/[END] blocks to MetaHackathon inference

Updated Docker FIle

Updated Inference.py

Add uv.lock and server/app.py

Add pyproject.toml for package management

Fix: make /reset endpoint accept empty request bodies

Fix: use HF Spaces port 7860 instead of 8000

Use explicit, clean multi-stage Dockerfile for frontend + backend

Simplify health endpoint - remove response_model validation

Fix: make health check and startup non-blocking

Add all missing directories and config files: components, hooks, styles, scripts, nextjs configs

Fix: correct import path for email_triage_env module

Fix: lazy load sessions and use uvicorn startup event

Fix Dockerfile: copy entire .next directory instead of standalone

Force rebuild: use Node.js 20 for Next.js 16 compatibility

Add missing components and hooks directories

Add full app - frontend and backend together

Add proper README with HF Space metadata

Add OpenEnv email triage environment

initial commit