-
step4: word-level tally note
-
step4: clean re-run PASS, step5 ready
-
step4: dakshina skipped, clean re-run pending
-
step4: hinglish top-up approved
-
step4: clean PASS+WARNING, hinglish gap
-
plan tally 500k on-plan + policy line
-
step4: clean_dedup.py + pull status
-
docs: DATA_SOURCES FINAL with verified IDs
-
docs: Step4 data quality checklist
-
step3 CLOSE: lock 0.05 (yaml+json+code)
-
step3: G loss 3.8244 confirmed
-
step3: F+G PASS, 0.10 lock pending
-
step3: add F(0.10/50) + G(0.05/100)
-
step3: lock deferred again, open
-
step3: fair-50 C wins, lock pending
-
step3: fair-50 matrix D+E, lock logic
-
step3: lock deferred, best result awaited
-
step3: Config C Kaggle PASS, 0.05 lock pending
-
step3: toy add Config C (0.05, 50 steps)
-
step3: toy Kaggle PASS A+B, lock pending approval
-
docs: KAGGLE_RUN session-proof guide
-
step3: toy_router_test.py for Kaggle (no local run per policy)
-
docs: ROADMAP_SCRATCH Step 0-10 with status
-
policy: all tests on Kaggle only
-
tests: 5/5 code-phase suite PASS + reports
-
docs: TEST_REPORT + PROGRESS Kaggle smoke
-
fix: smoke fallback for Kaggle clone without data/raw
-
docs: rename + push log
-
init: 28L MoE 241M audit-fixed, smoke pass
-
initial commit
