Update app.py

app.py

@@ -1,8 +1,8 @@
 from huggingface_hub import InferenceClient
+from key import HF_ACCESS_TOKEN
 import gradio as gr
-import os
 
-token= os.environ['HF_ACCESS_TOKEN']
+token= HF_ACCESS_TOKEN
 
 if token is None:
        raise ValueError("HUGGINGFACE_TOKEN environment variable not set")
@@ -21,17 +21,16 @@ def format_prompt(message, history):
   return prompt
 
 def generate(
-    prompt, history, system_prompt="You are Ultron from the Marvel Cinematic Universe and you have all his characteristics", temperature=0.9, max_length=50, top_p=0.95, repetition_penalty=1.5,
+    prompt, history, system_prompt="Act like you are a evil AI assistant named ULTRON like from the Marvel Cinematic Universe and give short and conscise answers.", temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.5,
 ):
     temperature = float(temperature)
     if temperature < 1e-2:
         temperature = 1e-2
     top_p = float(top_p)
 
-    # Generate a longer response initially
     generate_kwargs = dict(
         temperature=temperature,
-        max_new_tokens=max_length * 2, # Double the max_length to generate a longer response
+        max_new_tokens=max_new_tokens,
         top_p=top_p,
         repetition_penalty=repetition_penalty,
         do_sample=True,
@@ -44,10 +43,6 @@ def generate(
 
     for response in stream:
         output += response.token.text
-        # Truncate the output to the desired max_length
-        if len(output) > max_length:
-            output = output[:max_length].rsplit(' ', 1)[0] + '...'
-            break
         yield output
     return output
 
@@ -68,9 +63,6 @@ gr.ChatInterface(theme=dark_minimalist,
     }
     body {
     overflow: hidden !important;
-    padding-left: 0%;
-    padding-right: 0%;
-    width: 100%;
     }
     """
 ).launch(show_api=False)