Update app.py

app.py

@@ -26,19 +26,19 @@ def query_model(model_name: str, messages: List[Dict[str, str]]) -> Generator[st
     
     # Model-specific prompt formatting with full history
     model_prompts = {
-    "Qwen2.5-Coder-32B-Instruct": (
-        f"<|im_start|>system\nProvide detailed technical analysis of:\n{conversation}<|im_end|>\n"
-        "<|im_start|>assistant\nTechnical perspective:"
-    ),
-    "Qwen2.5-72B-Instruct": (
-        f"<|im_start|>system\nBuild upon this technical analysis:\n{conversation}<|im_end|>\n"
-        "<|im_start|>assistant\nCritical analysis:"
-    ),
-    "Llama3.3-70B-Instruct": (
-        "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n"
-        f"Synthesize and refine this analysis:\n{conversation}<|eot_id|>\n"
-        "<|start_header_id|>assistant<|end_header_id|>\nIntegrated perspective:"
-    )
+        "Qwen2.5-72B-Instruct": (
+            f"<|im_start|>system\nCollaborate with other experts. Previous discussion:\n{conversation}<|im_end|>\n"
+            "<|im_start|>assistant\nMy analysis:"
+        ),
+        "Llama3.3-70B-Instruct": (
+            "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n"
+            f"Build upon this discussion:\n{conversation}<|eot_id|>\n"
+            "<|start_header_id|>assistant<|end_header_id|>\nMy contribution:"
+        ),
+        "Qwen2.5-Coder-32B-Instruct": (
+            f"<|im_start|>system\nTechnical discussion context:\n{conversation}<|im_end|>\n"
+            "<|im_start|>assistant\nTechnical perspective:"
+        )
     }
 
     client = InferenceClient(base_url=endpoint, token=HF_API_KEY)