Update app.py

app.py

@@ -16,10 +16,20 @@ import os
 
 def http_bot_yield(prompt):
     headers = {"User-Agent": "vLLM Client"}
+    system_message = """
+        Below is an instruction that describes a task. Write a response that appropriately completes the request.
+        """
+    # pload = {
+    #     "prompt": f"<s>[INST] {prompt} [/INST] ",
+    #     "stream": True,
+    #     "max_tokens": 1024,
+    #     "temperature": 0.1
+    # }
+
     pload = {
-        "prompt": f"<s>[INST] {prompt} [/INST] ",
+        "prompt": f"### System:\n{system_message}\n\n\n\n### Instruction:\n{prompt}\n\n### Response:\n",
         "stream": True,
-        "max_tokens": 1024,
+        "max_tokens": 2048,
         "temperature": 0.1
     }
     response = requests.post(