from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "NousResearch/Hermes-2-Pro-Llama-3-8B"

model = None
tokenizer = None

# Example messages input
# messages = [
# {"role": "system", "content": "You are Hermes 2."},
# {"role": "user", "content": "Hello, who are you?"}
#]

def load():
  global model
  global tokenizer
  
  model = AutoModelForCausalLM.from_pretrained(model_name)
  tokenizer = AutoTokenizer.from_pretrained(model_name)
  
def generate(messages):
  gen_input = tokenizer.apply_chat_template(messages, return_tensors="pt")
  output_ids = model.generate(**gen_input, num_beams=5, no_repeat_ngram_size=2)
  
  return tokenizer.decode(output_ids[0], skip_special_tokens=True)