Spaces:
Sleeping
Sleeping
Commit History
-
Final Changes
344c3c9 -
Final Changes
71d64d0 -
New Final
1939cbc -
hf new add
7a23e48
-
Enhance Code Review Environment with Action History, Valid Actions, and Improved Grading
Harshit2N- Added action history tracking in CodeReviewEnv to store recent actions. - Implemented valid_actions method to return available actions based on the current state. - Updated reset method to accept a seed for randomization. - Improved step method to handle action processing and state completion more robustly. - Enhanced TaskGrader with new grading metrics for false positives and efficiency. - Updated diagnostics to include efficiency bonus and false positive penalties. - Added render and summary methods in CodeReviewEnv for better visualization and reporting. - Refactored inference.py to support batch processing of tasks and improved output handling. - Added difficulty levels to tasks in TaskDefinitions for better task categorization.5e92b80 -
Cleanup repo artifacts and finalize phase-1 environment updates
8c51832
-
Normalize provider-agnostic API config and add Space server endpoints
Shardul Dhekaneedaad73
-
CRE Main Update
Shardul DhekaneIntroduce a full code review environment and related assets: add environment package (env, models, tasks, graders, init) implementing CodeReviewEnv, Pydantic models, task definitions and grading/reward logic. Add multiple task JSON fixtures and simple data context modules (counter, file_handler, statistics, string_builder, user_repository) and tests (tests/test_env.py) to validate environment behavior. Update inference.py to use the OpenAI client and prefer HF_TOKEN/OPENAI_API_KEY for auth, and simplify chat completion. Adjust Dockerfile and docker-compose.yml to require HF_TOKEN (preferred) and update default task IDs/commands. Update openenv.yaml to reflect new task ids and API key requirements. Add project dependencies to requirements.txt.a21d9ae