Normalize entity_to_token schema (0..19) and document split groups; update human_train, human_test, synthetic_train

README.md

@@ -49,6 +49,15 @@ Each example includes: the NLQ, database identifier, a canonical dataset id, the
 ---
 
 ## Splits
+
+### Split groups
+
+- Human
+  - Human_train (`human_train`)
+  - Human_test (`human_test`)
+- Synthetic
+  - Synthetic_train (`synthetic_train`)
+
 **Entity token prevalence is consistent across splits: ~29% entity vs. ~71% `O`.**
 
 | Split             | # Examples |

synthetic_train.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1c14edb31b224bd44eaa4e5a71be828973c00f8f1f6d2e3e01fdf9d0c964651
-size 14669173
+oid sha256:1e06a98bc289b05fb179d7bb5857cc28746e0fa9755e5c9464b7cd266f5450a3
+size 17048559