Round 2 Implementation: Multi-period campaign, regulatory shocks, adversarial grading, and GRPO training infrastructure 52f5c27 balloonmann commited on Apr 22
Harden score constraints in app and align inference Round-1 defaults 4943c0a balloonmann commited on Apr 8
Scrub remaining instances of raw 0.0 and 1.0 logic to 0.01 and 0.99 a47adf4 balloonmann commited on Apr 8
Fix inference stdout formatting and update graders tests for strictly clamped reward constraint bounds 21fc032 balloonmann commited on Apr 7
fix: global invariant enforcement 0<score<1 on api pit stop and robust tests a16cc4e balloonmann commited on Apr 7
fix: clamp scores to (0,1) exclusive for Phase 2 validation; add root requirements.txt; defensive imports in inference.py 09ef9e7 balloonmann commited on Apr 7
fix: resolve README merge conflicts + add mandatory [START]/[STEP]/[END] structured logging to inference.py a7257f6 balloonmann commited on Apr 3