Add the LibriBrain holdout score

README.md

@@ -11,6 +11,23 @@ datasets:
 metrics:
 - f1
 library_name: pytorch
+
+model-index:
+- name: megconformer-speech-detection
+  results:
+  - task:
+      type: audio-classification
+      name: Speech classification
+    dataset:
+      name: LibriBrain 2025 PNPL (Standard track, speech task)
+      type: pnpl/LibriBrain
+      split: holdout
+    metrics:
+    - name: F1-macro
+      type: f1
+      value: 0.8890   # 88.90 %
+      args:
+        average: macro
 ---
 
 # MEGConformer for Speech Detection