import os

from transformers import AutoModelForCausalLM, AutoTokenizer

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
]

device = "cuda"  # the device to load the model onto
model_path = os.path.dirname(__file__)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_path)
response = ''
if __name__ == '__main__':

    while True:
        # prompt = "Give me a short introduction to large language model."
        prompt = input("input:")
        messages.append({"role": "user", "content": prompt})
        text = tokenizer.apply_chat_template(
            messages,
            tokenize=False,
            add_generation_prompt=True
        )
        model_inputs = tokenizer([text], return_tensors="pt").to(device)

        generated_ids = model.generate(
            model_inputs.input_ids,
            max_new_tokens=512
        )
        generated_ids = [
            output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
        ]

        response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
        print(response)
        messages.append({"role": "system", "content": response}, )