Update app.py

app.py

@@ -6,9 +6,6 @@ peft_model_id = f"Bsbell21/llm_instruction_generator"
 model = AutoModelForCausalLM.from_pretrained(peft_model_id, return_dict=True, device_map='auto')
 # tokenizer = AutoTokenizer.from_pretrained(peft_model_id)
 mixtral_tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x7B-v0.1")
-# merged_model = model.merge_and_unload()
-# Load the Lora model
-# model = PeftModel.from_pretrained(model, peft_model_id)
 
 def input_from_text(text):
   return "<s>[INST]Use the provided input to create an instruction that could have been used to generate the response with an LLM.\n" + text + "[/INST]"
@@ -21,9 +18,8 @@ def get_instruction(text):
       max_new_tokens=150,
       generation_kwargs={"repetition_penalty" : 1.7}
   )
-  # print(mixtral_tokenizer.decode(outputs[0], skip_special_tokens=True))
   print(mixtral_tokenizer.decode(outputs[0], skip_special_tokens=True))
-  return mixtral_tokenizer.decode(outputs[0], skip_special_tokens=True)
+  return mixtral_tokenizer.decode(outputs[0], skip_special_tokens=True).split("[/INST]")[1]
 
 if __name__ == "__main__":
     # make a gradio interface