getting local to work to prevent burining up my credits

.gitignore

@@ -1,4 +1,4 @@
-venv
-.venv
+venv/
+.venv/
 __pycache__
-gpt2-medium
+models/

requirements.txt

@@ -3,12 +3,13 @@ gradio[mcp]
 gradio
 
 # core Llama-Index + HF model support
-llama-index>=0.9.0  # Specify minimum version
-llama-index-embeddings-huggingface
 openai
-transformers
 torch
+transformers[torch]
 accelerate
+llama-index>=0.9.0  # Specify minimum version
+llama-index-embeddings-huggingface
+llama-index-llms-huggingface
 
 # optional extras
 langchain

src/app.py

@@ -1,12 +1,7 @@
 import os
 import gradio as gr
-from llama_index.core import VectorStoreIndex
-from llama_index.llms import HuggingFaceLLMPredictor
-from llama_index.readers import SimpleDirectoryReader
-
-# Relative imports should be explicit
-from parse_tabular import symptom_index  # Changed from relative import
-from ..utils.llama_index_utils import get_llm_predictor, build_index, query_symptoms
+from llama_index.llms.huggingface import HuggingFaceLLM
+from parse_tabular import symptom_index
 
 # --- System prompt ---
 SYSTEM_PROMPT = """
@@ -22,7 +17,7 @@ def process_speech(new_transcript, history):
         return history
     
     # Build LLM predictor
-    llm_predictor = HuggingFaceLLMPredictor(model_name_or_path=os.getenv("HF_MODEL", "gpt2-medium"))
+    llm_predictor = HuggingFaceLLM(model_name=os.getenv("HF_MODEL", "models/gpt2-medium"))
     
     # Query index with conversation
     prompt = "\n".join([f"{role}: {msg}" for role, msg in history])
@@ -33,7 +28,7 @@ def process_speech(new_transcript, history):
     ).query(prompt)
     
     # Append the new exchange to history
-    history.append((new_transcript, response.response))
+    history.append((new_transcript, str(response)))
     return history
 
 # Build Gradio interface
@@ -41,7 +36,7 @@ demo = gr.Blocks()
 with demo:
     gr.Markdown("# Symptom to ICD-10 Code Lookup (Audio Input)")
     chatbot = gr.Chatbot(label="Conversation")
-    audio = gr.Audio(source="microphone", type="text", streaming=True)
+    audio = gr.Audio(type="filepath", streaming=True)
     
     audio.stream(
         process_speech,

src/parse_tabular.py

@@ -84,3 +84,21 @@ if __name__ == "__main__":
     else:
         main()  # Use default path
     symptom_index = create_symptom_index()
+    
+    # Test multiple queries
+    test_queries = [
+        "persistent cough with fever",
+        "severe headache with nausea",
+        "lower back pain",
+        "difficulty breathing"
+    ]
+    
+    print("\nTesting symptom matching:")
+    print("-" * 50)
+    
+    for query in test_queries:
+        response = symptom_index.as_query_engine().query(query)
+        print(f"\nQuery: {query}")
+        print(f"Relevant ICD-10 codes:")
+        print(str(response))
+        print("-" * 50)