Upload L2_English_Codemix.csv

SAMBANDH-L2 is a compositional cultural reasoning dataset focused on intersection-based multi-hop reasoning over Indian cultural knowledge. Each question requires identifying the Indian state or union territory associated with multiple cultural entities simultaneously, using both same-category and cross-category associations. The dataset contains both English and region-aware code-mixed variants spanning multiple Indian languages, enabling evaluation of compositional reasoning and multilingual robustness under linguistic variation.

L2_English_Codemix.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14a03b94eeed12fb55f66f276acd67add12e4492e4d731c2fd12f3219a4d6a76
+size 12209849