Update app.py

app.py

@@ -186,21 +186,33 @@ def get_default_system_prompt(personalized_data):
     """
     print(DEFAULT_SYSTEM_PROMPT)
     return DEFAULT_SYSTEM_PROMPT
-    
+
+## trust-game-llama-2-7b-chat
+# app.py
+def construct_input_prompt(chat_history, message, personalized_data):
+    input_prompt = f"<s>[INST] <<SYS>>\n{get_default_system_prompt(personalized_data)}\n<</SYS>>\n\n "
+    for user, assistant in chat_history:
+        input_prompt += f"{user} [/INST] {assistant} <s>[INST] "
+    input_prompt += f"{message} [/INST] "
+    return input_prompt
+
+## trust-game-llama-2-7b-chat
+# app.py 
 @spaces.GPU
 def generate(
-    request: gr.Request, # To fetch query params 
     message: str,
     chat_history: list[tuple[str, str]],
-    input_prompt: str,
+    # input_prompt: str,
     max_new_tokens: int = 1024,
     temperature: float = 0.6,
     top_p: float = 0.9,
     top_k: int = 50,
     repetition_penalty: float = 1.2,
-) -> Iterator[str]: 
-    conversation = []
+    request: gr.Request, # To fetch query params
+) -> Iterator[str]: # Change return type hint to Iterator[str]
 
+    conversation = []
+    
     # Fetch query params 
     params = request.query_params
     print('those are the query params')
@@ -214,30 +226,21 @@ def generate(
     if session_index is not None:
         print("Session index:", session_index)
     else:
-        print("Session index parameter not found or has no value.")
+        print("Session index not found or has no value.")
 
+    # Fetch personalized data
     personalized_data = fetch_personalized_data(session_index)
-    DEFAULT_SYSTEM_PROMPT = get_default_system_prompt(personalized_data)
-
-    ## trust-game-llama-2-7b-chat
-    # app.py
-    def construct_input_prompt(chat_history, message):
-        input_prompt = f"<s>[INST] <<SYS>>\n{DEFAULT_SYSTEM_PROMPT}\n<</SYS>>\n\n "
-        for user, assistant in chat_history:
-            input_prompt += f"{user} [/INST] {assistant} <s>[INST] "
-        input_prompt += f"{message} [/INST] "
-        return input_prompt
-    
-    # Construct the input prompt using the functions from the construct_input_prompt function 
-    input_prompt = construct_input_prompt(chat_history, message)
     
+    # Construct the input prompt using the functions from the system_prompt_config module
+    input_prompt = construct_input_prompt(chat_history, message, personalized_data)
+
     # Move the condition here after the assignment
     if input_prompt:
         conversation.append({"role": "system", "content": input_prompt})
 
     # Convert input prompt to tensor
     input_ids = tokenizer(input_prompt, return_tensors="pt").to(model.device)
-    
+
     for user, assistant in chat_history:
         conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
     conversation.append({"role": "user", "content": message})
@@ -249,10 +252,8 @@ def generate(
     input_ids = input_ids.to(model.device)
 
     # Set up the TextIteratorStreamer
-    #streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
-    # Set new timeout error mitigation
-    streamer = TextIteratorStreamer(tokenizer, timeout=30.0, skip_prompt=True, skip_special_tokens=True)
-
+    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
+    
     # Set up the generation arguments
     generate_kwargs = dict(
         {"input_ids": input_ids},
@@ -275,7 +276,6 @@ def generate(
     for text in streamer:
         outputs.append(text)
         yield "".join(outputs)
-        
 
 chat_interface = gr.ChatInterface(
 fn=generate,