preload

ai_med_extract/api/routes.py

@@ -319,6 +319,18 @@ def register_routes(app, agents):
     from ai_med_extract.utils.openvino_summarizer_utils import (
         parse_ehr_chartsummarydtl, visits_sorted, compute_deltas, build_compact_baseline, delta_to_text, build_main_prompt
     )
+        # ============ HF SPACES: PRE-LOAD GGUF MODEL AT STARTUP ============
+    print("[HF SPACES] ⏳ Pre-loading GGUF model to prevent startup timeout...")
+    try:
+        from ai_med_extract.utils.model_loader_gguf import GGUFModelPipeline
+        model_name = "microsoft/Phi-3-mini-4k-instruct-gguf"
+        filename = "Phi-3-mini-4k-instruct-q4.gguf"
+        # This will download and load the model NOW, during app startup
+        pipeline = GGUFModelPipeline(model_name, filename)
+        print("[HF SPACES] ✅ GGUF model pre-loaded successfully!")
+    except Exception as e:
+        print(f"[HF SPACES] ❌ Pre-load failed (fallback will handle it): {e}")
+    # ============ END PRE-LOAD ============
 
     @app.route('/api/patient_summary_openvino', methods=['POST'])
     def patient_summary_openvino():

requirements.txt

@@ -164,6 +164,7 @@ wrapt==1.17.3
 xxhash==3.5.0
 yarl==1.20.1
 llama-cpp-python==0.2.72
+ctransformers>=0.2.0
 
 # Add timeout and signal handling dependencies
 timeout-decorator==0.5.0