Add correct_decision field for eval success metric

Fix Dockerfile casing for HF build detection

Add HF Space Docker metadata

improved ipynb

Rename kaggle-grpo-train-ipynb.ipynb to kaggle-grpo-train.ipynb

uploaded kaggle trainning notebook

Update training notebook link to Kaggle

Implement JSON parsing and error handling in inference

Refine README formatting and content clarity

Apply manual fixes for GRPO training and docs

checkout: temporary commit for worktree checkout

checkout: temporary commit for worktree checkout

Update README.md

updated grpo logic

added grpo colab file

fixed docker

updated requirements

updated requirements

fixed docker

logic update 2.0

improved logic

updated docker

added demo

updated test file

updated src/models.py · src/environment.py · src/generator.py · apps/crm_api.py · apps/regulatory_api.py · apps/audit_api.py

Update readme

Update readme

Update readme

Fix task_id kwarg in reward function

Fix task_id kwarg in reward function

round 2 improvement updated GRPO

round 2 improvement updated GRPO

Phase 2 complete: Fixed inference loop and added phase gates

Update README

Clamp scores strictly between 0.01 and 0.99 for Scaler grader

Update inference.py

add main function

Fix Dockerfile casing and update server path

Fix OpenEnv structure for Scaler validation

Initial commit: Meta Ad-Policy Sandbox