normalize filtered probs (minor)

interfaces/cap_minor_media.py

@@ -39,6 +39,18 @@ for code in CAP_MIN_CODES:
     major_to_minor_map[major_id].append(code)
 major_to_minor_map = dict(major_to_minor_map)
 
+def normalize_probs(probs: dict):
+    if not probs:
+        return {}
+
+    min_val = min(probs.values())
+    max_val = max(probs.values())
+    range_val = max_val - min_val
+
+    if range_val == 0:
+        return {k: 1.0 for k in probs}
+
+    return {k: (v - min_val) / range_val for k, v in probs.items()}
 
 def check_huggingface_path(checkpoint_path: str):
     try:
@@ -102,6 +114,8 @@ def predict(text, major_model_id, minor_model_id, tokenizer_id, HF_TOKEN=None):
         # Restrict to valid minor codes
         valid_indices = [minor_id_to_index[mid] for mid in valid_minor_ids if mid in minor_id_to_index]
         filtered_probs = {minor_index_to_id[i]: float(minor_probs[0][i]) for i in valid_indices}
+        filtered_probs = normalize_probs(filtered_probs)
+        
         output_pred = {
             f"[{k}] {CAP_MIN_LABEL_NAMES[k]}": v
             for k, v in sorted(filtered_probs.items(), key=lambda item: item[1], reverse=True)