Update README for dataset/benchmark top-level organization

README.md

@@ -1,12 +1,6 @@
-# V1 Cognition Benchmark (Minimal)
+# UserModeling
 
-This minimal release contains only the model-ready training table and its basic stats.
-
-## Files
-- `cognition_wallet_market_labels_holdpos_noconf_mk10plus.parquet` (main training set)
-- `cognition_wallet_market_labels_holdpos_noconf_mk10plus_stats.csv` (quick dataset stats)
-- `SHA256SUMS.txt` (integrity checks)
-
-## Headline
-- wallets: 31,319
-- wallet×market rows: 1,810,609
+Canonical layout:
+- dataset/
+- benchmark/
+- metadata/