openenv.yaml

name: DebugArena
description: RL environment where LLMs learn to fix buggy Python code through trial and error
version: 1.0.0
theme: self-improvement
author: Bharath Vikas Tadepalli
base_url: https://bharathvikas-debugarena.hf.space
endpoints:
  reset:
    method: POST
    path: /reset
  step:
    method: POST
    path: /step
  state:
    method: GET
    path: /state
observation_space:
  buggy_code: string
  error_message: string
  test_results: list
  tests_passed: int
  tests_total: int
  attempts_remaining: int
  solved: bool
  feedback: string
action_space:
  fixed_code: string
  explanation: string
reward_range: [-1.3, 3.3]