Update app.py

app.py

@@ -108,10 +108,10 @@ def run_gpt(
         safe_search=safe_search,
     ) + prompt_template.format(**prompt_kwargs)
     if VERBOSE:
-        logging.info(LOG_PROMPT.format(content))  # Log the prompt 
+        logging.info(LOG_PROMPT.format(content=content))  # Log the prompt 
     resp = client.text_generation(content, max_new_tokens=max_new_tokens, stop_sequences=stop_tokens, temperature=0.7, top_p=0.8, repetition_penalty=1.5)
     if VERBOSE:
-        logging.info(LOG_RESPONSE.format([resp]))  # Log the response
+        logging.info(LOG_RESPONSE.format(resp=resp))  # Log the response
     return resp
 
 def generate(
@@ -121,15 +121,14 @@ def generate(
     logging.info(f"Seed: {seed}")  # Log the seed
 
     # Set the agent prompt based on agent_name
+    agent = "You are a helpful AI assistant."
     if agent_name == "WEB_DEV":
-        agent = "You are a helpful AI assistant. You are a web developer."
+        agent += " You are a web developer."
     elif agent_name == "AI_SYSTEM_PROMPT":
-        agent = "You are a helpful AI assistant. You are an AI system."
+        agent += " You are an AI system."
     elif agent_name == "PYTHON_CODE_DEV":
-        agent = "You are a helpful AI assistant. You are a Python code developer."
-    else:
-        agent = "You are a helpful AI assistant."
-
+        agent += " You are a Python code developer."
+    
     system_prompt = f"{agent} {sys_prompt}".strip()
 
     temperature = max(float(temperature), 1e-2)
@@ -142,9 +141,10 @@ def generate(
     formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5)  # Truncated history
     logging.info(f"Formatted Prompt: {formatted_prompt}")
 
-    client = InferenceClient(model) if model != "mistralai/Mixtral-8x7B-Instruct-v0.1" else InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
+    # Conditionally create client
+    this_client = InferenceClient(model) if model != "mistralai/Mixtral-8x7B-Instruct-v0.1" else client
 
-    stream = client.text_generation(
+    stream = this_client.text_generation(
         formatted_prompt,
         temperature=temperature,
         max_new_tokens=max_new_tokens,
@@ -329,27 +329,31 @@ def format_prompt(message, history, max_history_turns=5):
     prompt += f" {bot_response}</s> "
   prompt += f"[INST] {message} [/INST]"
   return prompt
+
 agents =[
     "WEB_DEV",
     "AI_SYSTEM_PROMPT",
     "PYTHON_CODE_DEV"
 ]
+
 def generate(
     prompt, history, agent_name=agents[0], sys_prompt="", temperature=0.9, max_new_tokens=2048, top_p=0.95, repetition_penalty=1.0, model="mistralai/Mixtral-8x7B-Instruct-v0.1"
 ):
     seed = random.randint(1,1111111111111111)
+    logging.info(f"Seed: {seed}")  # Log the seed
 
-    # Correct the line:
+    # Set the agent prompt based on agent_name
+    agent = "You are a helpful AI assistant."
     if agent_name == "WEB_DEV":
-        agent = "You are a helpful AI assistant. You are a web developer."
-    if agent_name == "AI_SYSTEM_PROMPT":
-        agent = "You are a helpful AI assistant. You are an AI system."
-    if agent_name == "PYTHON_CODE_DEV":
-        agent = "You are a helpful AI assistant. You are a Python code developer."
-    system_prompt = agent
-    temperature = float(temperature)
-    if temperature < 1e-2:
-        temperature = 1e-2
+        agent += " You are a web developer."
+    elif agent_name == "AI_SYSTEM_PROMPT":
+        agent += " You are an AI system."
+    elif agent_name == "PYTHON_CODE_DEV":
+        agent += " You are a Python code developer."
+    
+    system_prompt = f"{agent} {sys_prompt}".strip()
+
+    temperature = max(float(temperature), 1e-2)
     top_p = float(top_p)
 
     # Add the system prompt to the beginning of the prompt
@@ -358,14 +362,28 @@ def generate(
     # Use 'prompt' here instead of 'message'
     formatted_prompt = format_prompt(formatted_prompt, history, max_history_turns=5)  # Truncated history
     logging.info(f"Formatted Prompt: {formatted_prompt}")
-    stream = client.text_generation(formatted_prompt, temperature=temperature, max_new_tokens=max_new_tokens, top_p=top_p, repetition_penalty=repetition_penalty, stream=True, details=True, return_full_text=False)
-resp = ""
+
+    # Conditionally create client
+    this_client = InferenceClient(model) if model != "mistralai/Mixtral-8x7B-Instruct-v0.1" else client
+
+    stream = this_client.text_generation(
+        formatted_prompt,
+        temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_p=top_p,
+        repetition_penalty=repetition_penalty,
+        stream=True,
+        details=True,
+        return_full_text=False
+    )
+    resp = ""
     for response in stream:
         resp += response.token.text
         yield resp  # This allows for streaming the response
 
     if VERBOSE:
-        logging.info(LOG_RESPONSE.format(resp))  # Pass resp to format
+        logging.info(f"RESPONSE: {resp}")  # Log the response directly
+
 
 def generate_text_chunked(input_text, model, generation_parameters, max_tokens_to_generate):
     """Generates text in chunks to avoid token limit errors."""
@@ -387,17 +405,7 @@ def generate_text_chunked(input_text, model, generation_parameters, max_tokens_t
 
     return ''.join(generated_text)
 
-    formatted_prompt = format_prompt(prompt, history, max_history_turns=5)  # Truncated history
-    logging.info(f"Formatted Prompt: {formatted_prompt}")
-    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
-    output = ""
-
-    for response in stream:
-        output += response.token.text
-        yield output
-    return output
-
-
+    
 additional_inputs=[
     gr.Dropdown(
         label="Agents",