Upload folder using huggingface_hub

Dockerfile

@@ -11,7 +11,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     && rm -rf /var/lib/apt/lists/*
 
 # Install Python dependencies
-COPY server/requirements.txt .
+COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
 # Copy environment code
@@ -25,8 +25,8 @@ ENV PYTHONPATH="/app:$PYTHONPATH"
 
 # Health check
 HEALTHCHECK --interval=30s --timeout=3s --start-period=5s --retries=3 \
-    CMD curl -f http://localhost:8002/health || exit 1
+    CMD curl -f http://localhost:7860/health || exit 1
 
 # Run the FastAPI server
 ENV ENABLE_WEB_INTERFACE=true
-CMD ["uvicorn", "server.app:app", "--host", "0.0.0.0", "--port", "8002"]
+CMD ["uvicorn", "server.app:app", "--host", "0.0.0.0", "--port", "7860"]

PROJECT_SUMMARY.md

@@ -0,0 +1,43 @@
+# Project Summary: MedTriage OpenEnv
+**Meta PyTorch OpenEnv Hackathon Submission**
+
+## 🏥 Project Overview
+**MedTriage** is a real-world simulation environment designed for training and evaluating AI agents in clinical decision-making. Unlike "toy" environments, it models a genuine human task: categorizing patient symptoms into appropriate care levels (Self-Care, Clinic, Urgent Care, or Emergency).
+
+## 🛠️ What We Built
+We have developed a complete, spec-compliant OpenEnv environment with the following components:
+
+### 1. Core Environment (`server/triage_environment.py`)
+- **Action Space**: A `triage_patient` tool where agents assign a triage level (0-3) and provide clinical reasoning.
+- **Observation Space**: Detailed patient profiles including demographics, unstructured symptom text, vitals (BP, Temp, HR, SpO2), and medical history.
+- **Reward Function**: A medical-safety-first scoring system (0.0 - 1.0) that rewards accuracy and penalizes dangerous under-triage (e.g., missing an emergency).
+
+### 2. Mandatory Tasks
+Implemented three specific scenarios with automated graders:
+- **TASK_EASY**: Seasonal Allergies (Self-Care)
+- **TASK_MEDIUM**: Possible Appendicitis (Urgent Care)
+- **TASK_HARD**: Atypical Myocardial Infarction in elderly patient (Emergency)
+
+### 3. API Infrastructure (`server/app.py`)
+- Implemented standard OpenEnv endpoints: `/reset`, `/step`, `/state`.
+- Added required Hackathon endpoints:
+    - `/tasks`: Dynamic list of available scenarios and schemas.
+    - `/grader`: Real-time performance scoring.
+    - `/baseline`: Automated trigger for the inference script.
+
+### 4. Baseline Inference (`inference.py`)
+- Created a reproducible baseline script that uses medical heuristics to evaluate the environment. It ensures the environment is "solveable" and provides a benchmark score for comparison.
+
+## 🚀 Technical Improvements & Fixes
+To ensure the project meets the highest submission standards, we performed the following:
+- **Spec Compliance**: Added missing `__init__.py` files and standardized the directory structure to pass `openenv validate`.
+- **Dependency Management**: Updated `pyproject.toml` and `server/requirements.txt` to include `openenv-core>=0.2.0` and generated a `uv.lock` file.
+- **Containerization**: Optimized the `Dockerfile` to support root-level builds, enabled health checks, and added editable installation (`pip install -e .`) to resolve local module imports.
+- **Deployment Automation**: Configured `README.md` with Hugging Face Space metadata (`app_port: 8002`, `sdk: docker`) for seamless hosting.
+
+## 🌐 Deployment Links
+- **GitHub Repository**: [https://github.com/ash399/med-triage-openenv](https://github.com/ash399/med-triage-openenv)
+- **Hugging Face Space**: [https://huggingface.co/spaces/ashdev/med-triage-openenv](https://huggingface.co/spaces/ashdev/med-triage-openenv)
+
+---
+*Documented on: Monday, March 30, 2026*

README.md

@@ -3,7 +3,7 @@ title: MedTriage OpenEnv
 emoji: 🏥
 sdk: docker
 pinned: false
-app_port: 8002
+app_port: 7860
 tags:
   - openenv
   - healthcare

requirements.txt

@@ -0,0 +1,6 @@
+fastapi
+uvicorn
+pydantic
+fastmcp
+requests
+openenv-core>=0.2.0

server/app.py

@@ -68,7 +68,7 @@ async def trigger_baseline():
         from inference import run_baseline
         
     # Execute actual baseline
-    scores = run_baseline(base_url="http://localhost:8002")
+    scores = run_baseline(base_url="http://localhost:7860")
     
     return {
         "status": "baseline_completed",
@@ -77,7 +77,7 @@ async def trigger_baseline():
 
 def main():
     import uvicorn
-    uvicorn.run(app, host="0.0.0.0", port=8002)
+    uvicorn.run(app, host="0.0.0.0", port=7860)
 
 if __name__ == "__main__":
     main()