Create eval.yaml

eval.yaml

@@ -0,0 +1,11 @@
+name: FINAL Bench
+description: >
+  Measuring Functional Metacognitive Reasoning in Large Language Models.
+  100 expert-level tasks with hidden cognitive traps across 15 domains
+  and 8 TICOS metacognitive types.
+evaluation_framework: final-bench
+
+tasks:
+  - id: metacognitive
+    config: default
+    split: train