Sync latest app + data (clinician groups, jsonl)

chatgpt.py

@@ -786,11 +786,16 @@ with gr.Blocks() as app:
     jsonl_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'Qwen__Qwen2.5-7B-Instruct_io.jsonl')
     jsonl_loaded = False
     eval_data_by_patient = {}  # Initialize empty to ensure clean start
+    
+    # FORCE: Only load from Qwen JSONL, completely skip all other data sources
     if os.path.exists(jsonl_path):
+        print(f"[INFO] ========== LOADING FROM QWEN JSONL ONLY ==========")
         print(f"Loading data from JSONL: {jsonl_path}")
         jsonl_data = load_jsonl(jsonl_path)
         print(f"Found {len(jsonl_data)} entries in JSONL file")
+        entry_count = 0
         for entry in jsonl_data:
+            entry_count += 1
             full_id = entry.get('patient_id', '') or ''
             # Extract patient_id from format like "11318742_admission_29646478_input"
             # Only keep the patient_id part (e.g., "11318742")
@@ -830,8 +835,11 @@ with gr.Blocks() as app:
                 'highlights': []
             }
         jsonl_loaded = True
-        print(f"Loaded {len(eval_data_by_patient)} patients from JSONL")
+        print(f"Processed {entry_count} entries from JSONL")
+        print(f"Loaded {len(eval_data_by_patient)} unique patients from JSONL")
         print(f"Patient IDs: {sorted(list(eval_data_by_patient.keys()))}")
+        print(f"[INFO] JSONL loading complete. jsonl_loaded = {jsonl_loaded}")
+        print(f"[INFO] ================================================")
     else:
         print(f"[WARNING] JSONL file not found: {jsonl_path}")
         print(f"[WARNING] Will attempt to load from fallback sources")
@@ -839,7 +847,9 @@ with gr.Blocks() as app:
     # Fallback to old format if JSONL not found
     # IMPORTANT: Only load from fallback if JSONL was NOT loaded
     if (not jsonl_loaded) and os.path.exists(summaries_dir):
+        print(f"[WARNING] ========== FALLBACK: Loading from summaries_dir ==========")
         print(f"[INFO] Loading from fallback summaries_dir: {summaries_dir}")
+        print(f"[WARNING] This should NOT happen if Qwen JSONL exists!")
         files = os.listdir(summaries_dir)
         pattern = re.compile(r"^subject_(\d+)_hadm_(\d+)_(model_input|summary)\.txt$")
         for fname in files:
@@ -889,8 +899,10 @@ with gr.Blocks() as app:
     # IMPORTANT: Only load if JSONL was NOT loaded (to avoid duplicate patients)
     per_adm_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'per_admission_results_30patient_2ad')
     if (not jsonl_loaded) and os.path.exists(per_adm_dir):
+        print(f"[WARNING] ========== FALLBACK: Loading from per_admission_results_30patient_2ad ==========")
         print(f"[WARNING] JSONL not loaded, falling back to per_admission_results_30patient_2ad directory")
         print(f"[WARNING] This should NOT happen if Qwen JSONL exists!")
+        print(f"[WARNING] jsonl_loaded flag = {jsonl_loaded}")
         for fname in os.listdir(per_adm_dir):
             if not fname.startswith('patient_') or not fname.endswith('.json') or fname == 'all_patients_per_admission.json':
                 continue