change

app.py

@@ -18,6 +18,8 @@ model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path,
                                              token = os.environ['TOKEN']
                                              )
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+if tokenizer.pad_token is None:
+    tokenizer.pad_token = tokenizer.eos_token
 
 @spaces.GPU
 def generate(prompt, temperature=0.7, top_k=50, repetition_penalty=1.2):
@@ -30,7 +32,7 @@ def generate(prompt, temperature=0.7, top_k=50, repetition_penalty=1.2):
                         top_k=top_k)
     output=tokenizer.decode(output_ids[0],skip_special_tokens=True)
     # Remove the prompt from the output
-    output=output[len(formatted_prompt):]
+    #output=output[len(formatted_prompt):]
     return output
 
 prompt_input=gr.Textbox(