Adharshts/meta_pytorch_hackathon

Fix task structure: added 3 tasks and updated YAML for autograder

Fix namespace collision by renaming tasks to eval_tasks

Update grader structure to match expected OpenEnv package format

Rename grader file to fix namespace collision

Move graders to root directory to fix import dropping

Add __init__.py to make server a module and fix YAML schema

Fix yaml schema to prevent silent pydantic drops

Fix inference logs bounding and restore explicit YAML graders

Fix cumulative reward math trap to strictly bound score between 0.001 and 0.999

Isolate grader to fix namespace collision and import drops

Add explicit grader function and strict YAML schema

Fix phase 2 ranges and tasks

Finalize folder structure and fix internal server error

Fix FastAPI 500 error by adding truncated to Observation

Fix Internal Server Error by returning Observation in reset

Fix missing Observation import

Add Uvicorn server initialization to FastAPI app

Update python requirement and add uv.lock

Update Dockerfile uvicorn path to server.app:app

Fix folder structure and add uv.lock for grader

Add required pyproject.toml for multi-mode deployment

Add required openenv.yaml config file

Add inference.py to repo root for grader

Pass class instead of instance to create_fastapi_app

Fix FastAPI app initialization

Update server and env for Task 1: Disk Full Outage

Remove dataclass decorators to fix Pydantic conflict

Add missing test_env.py to unblock build

Pass class blueprint instead of instance

Add missing state method to OpenEnv adapter

Implement OpenEnv adapter and dataclasses

Instantiate env object

Implement actual OpenEnv server to handle POST requests

Remove missing test_env.py from Dockerfile

Add missing Hugging Face Space configuration

Fix HF Space health check timeout

Initial commit: Meta PyTorch Hackathon submission