Update app.py

app.py

@@ -10,7 +10,7 @@ def generate_text(prompt):
  input_ids = tokenizer.encode(prompt, return_tensors="pt")
  attention_mask = torch.ones(input_ids.shape, dtype=torch.bool)
 
- max_length = model.config.n_positions if len(input_ids[0]) > model.config.n_positions else len(input_ids[0]) + 50
+ max_length = model.config.n_positions if len(input_ids[0]) > model.config.n_positions else len(input_ids[0]) + 20
  beam_output = model.generate(
  input_ids,
  attention_mask=attention_mask,
@@ -20,8 +20,8 @@ def generate_text(prompt):
  no_repeat_ngram_size=2,
  early_stopping=True,
  temperature=0.9,
- top_p=0.95,
- top_k=70,
+ top_p=0.90,
+ top_k=50,
  length_penalty=2.0,
  do_sample=True,
  eos_token_id=tokenizer.eos_token_id,
@@ -30,22 +30,22 @@ def generate_text(prompt):
  
  text = tokenizer.decode(beam_output[0], skip_special_tokens=True)
  return text
-h1 = "Löwolf Chat"
+
 css = """
  body { font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; }
- .gradio_app { max-width: 600px; margin: 50px auto; }
+ .gradio_app { display: flex; flex-direction: column-reverse; max-width: 600px; margin: 50px auto; }
  .gradio_interface { box-shadow: 0 0 20px rgba(0,0,0,0.1); }
  .gradio_input_container { width: 100%; }
  .gradio_output_container { width: 100%; margin-bottom: 25px; }
- .gradio_input { border-radius: 20px; }
- .gradio_output { border-radius: 20px; }
+ .gradio_text_input { height: 50px; border-radius: 20px; }
+ .gradio_text_output { height: calc(100vh - 150px); border-radius: 20px; overflow-y: auto; }
  button { border-radius: 20px; }
 """
 
 iface = gr.Interface(
  fn=generate_text,
- inputs=gr.Textbox(placeholder="Type a message..."),
- outputs="text",
+ inputs=gr.Textbox(lines=2, placeholder="Type a message...", label="Your Message"),
+ outputs=gr.Textbox(label="Löwolf Chat Responses", placeholder="Responses will appear here...", interactive=False, lines=10),
  css=css
 )
 
@@ -53,3 +53,4 @@ iface.launch()
 
 
 
+