:clown_face: clown

knowledge_cutoff_demo.py

@@ -185,39 +185,43 @@ def generate_single_response(model_name, input_question):
 @spaces.GPU
 def process_all_models_parallel(input_question):
     """Process all models in parallel for maximum speed"""
-    if not input_question.strip():
-        return "❌ Please enter a question", "❌ Please enter a question", "❌ Please enter a question"
+    # if not input_question.strip():
+    #     return "❌ Please enter a question", "❌ Please enter a question", "❌ Please enter a question"
     
-    start_time = time.time()
+    # start_time = time.time()
     
-    # Use ThreadPoolExecutor for parallel processing
-    with ThreadPoolExecutor(max_workers=3) as executor:
-        # Submit all tasks simultaneously
-        futures = {
-            executor.submit(generate_single_response, model_name, input_question): model_name 
-            for model_name in MODEL_CONFIGS.keys()
-        }
+    # # Use ThreadPoolExecutor for parallel processing
+    # with ThreadPoolExecutor(max_workers=3) as executor:
+    #     # Submit all tasks simultaneously
+    #     futures = {
+    #         executor.submit(generate_single_response, model_name, input_question): model_name 
+    #         for model_name in MODEL_CONFIGS.keys()
+    #     }
         
-        # Collect results as they complete
-        results = {}
-        for future in futures:
-            model_name = futures[future]
-            try:
-                result = future.result(timeout=45)  # Longer timeout for Llama-1
-                results[model_name] = result
-            except Exception as e:
-                results[model_name] = f"❌ Timeout or error for {model_name}: {str(e)}"
+    #     # Collect results as they complete
+    #     results = {}
+    #     for future in futures:
+    #         model_name = futures[future]
+    #         try:
+    #             result = future.result(timeout=45)  # Longer timeout for Llama-1
+    #             results[model_name] = result
+    #         except Exception as e:
+    #             results[model_name] = f"❌ Timeout or error for {model_name}: {str(e)}"
     
-    total_time = time.time() - start_time
+    # total_time = time.time() - start_time
     
-    # Add total timing to first response  
-    llama1_response = results.get("Llama-1 7B", "❌ Error")
+    # # Add total timing to first response  
+    # llama1_response = results.get("Llama-1 7B", "❌ Error")
     
-    return (
-        llama1_response,
-        results.get("Llama-2 7B Chat", "❌ Error"),
-        results.get("Llama-3.2 3B", "❌ Error")
-    )
+    # return (
+    #     llama1_response,
+    #     results.get("Llama-2 7B Chat", "❌ Error"),
+    #     results.get("Llama-3.2 3B", "❌ Error")
+    # )
+    llama1_response = generate_single_response("Llama-1 7B", input_question)
+    llama2_response = generate_single_response("Llama-2 7B Chat", input_question)
+    llama3_response = generate_single_response("Llama-3.2 3B", input_question)
+    return llama1_response, llama2_response, llama3_response
 
 # def benchmark_models():
 #     """Benchmark all models with a test question"""