Update app.py

app.py

@@ -19,15 +19,12 @@ model = AutoModelForCausalLM.from_pretrained(model_path, device_map='cuda', quan
 
 @spaces.GPU
 def generate_text(usertitle, content, max_length, temperature):
-#   msg = [{
-#       'title': title,
-#       'content': content
-#   }]
     input_text = {'title': usertitle, 'content': content} 
     inputs = tokenizer.apply_chat_template(input_text, return_tensors='pt').cuda()
-    generated_text = tokenizer.decode(model.generate(inputs, max_new_tokens=max_length, temperature=temperature, skip_special_tokens=True, do_sample=True)[0])#.tokenizer.split(tokenizer.eos_token)[0]
-    
-    return generated_text
+    generated_text = tokenizer.decode(model.generate(inputs, max_new_tokens=max_length, temperature=temperature, do_sample=True)[0])#.strip().split(tokenizer.eos_token)[0]
+    split_text = generated_text.strip().split(tokenizer.eos_token)[0]
+
+    return split_text
 
 def gradio_app():
     with gr.Blocks() as demo: