Spaces:
Runtime error
Runtime error
Commit History
-
fix: resolve 500 error on /schema and add extra validation tasks
vineetshukla.work@gmail.com52fe477 -
fix: standardize grader structure to tasks/ directory for OpenEnv validation compliance
vineetshukla.work@gmail.coma594b6e -
fix: add mandatory /metadata and /schema endpoints for OpenEnv validation and update task metadata
vineetshukla.work@gmail.com5a87b28 -
Fix OpenEnv validation: Use explicit python grader for tasks
vineetshukla.work@gmail.com9bd4a93
-
fix: task routing by name, remove out-of-range rewards, add grader field to tasks
vineetshukla.work@gmail.comb64950c
-
fix: strictly bound task scores to satisfy OpenEnv Phase 2 testing bounds
vineetshukla.work@gmail.com5fcb94c -
fix: allow empty body for /reset endpoint to pass auto-grader
vineetshukla.work@gmail.com46525cf -
fix: rewrite inference.py & client.py to match OpenEnv sample pattern
vineetshukla.work@gmail.com- inference.py: async main, from_docker_image(), result.observation/done/reward, env.close() in finally - client.py: new CodeSenseiEnv with from_docker_image(), StepResult wrapper - env/__init__.py: export CodeSenseiEnv, CodeSenseiAction for clean imports"6cbd2ef -
feat: add inference.py, openenv.yaml, bug dataset & openai dep for hackathon submission
vineetshukla.work@gmail.com- inference.py: OpenAI client, [START]/[STEP]/[END] structured stdout logs, env var contract (API_BASE_URL, MODEL_NAME, HF_TOKEN) - openenv.yaml: environment spec for `openenv validate` compliance - env/data/bug_dataset.json: 10 Python bugs with tests - requirements.txt: added openai>=1.0.0 - PREVALIDATIONSCRIPT.md & sample-inference-script.py: reference docs"01620c1
-
feat: CodeSensei - GRPO-trained LLM code debugger on OpenEnv
vineetshukla.work@gmail.com- Custom OpenEnv environment with reset/step/state protocol - 5-signal reward system (correctness, progress, syntax, runtime, repetition) - Sandboxed Python execution with timeout and restricted imports - WebSocket + HTTP FastAPI server (HF Spaces compatible) - GRPO training script optimized for Colab T4 free tier - Gradio demo with baseline vs fine-tuned model comparison - 10 curated buggy Python functions with test cases - All integration tests passingc47c81c