import gradio as gr
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    pipeline,
    logging,
)
from peft import PeftModel


# Reload model in FP16 and merge it with LoRA weights
model = AutoModelForCausalLM.from_pretrained(
    'TinyLlama/TinyLlama-1.1B-Chat-v1.0'
)
model = PeftModel.from_pretrained(model, 'jezv/TinyLlama-1.1b-cmd')
model = model.merge_and_unload()

# Reload tokenizer to save it
tokenizer = AutoTokenizer.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0', trust_remote_code=True)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "right"


def greet(prompt):
    # Try question-answering
    pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, max_new_tokens=20)
    messages = [
        {
            "role": "system",
            "content": "You are a chatbot assistant who helps with writing console commands.",
        },
        {"role": "user", "content": prompt},
    ]
    text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    result = pipe(text)
    return result[0]['generated_text']

demo = gr.Interface(
    fn=greet,
    inputs=["text"],
    outputs=["text"],
)

demo.launch()