import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
from peft import PeftModel, PeftConfig


device = torch.device("cuda" if torch.cuda.is_available() else "cpu")


peft_model_id = "asusevski/mistraloo-sft"
peft_config = PeftConfig.from_pretrained(peft_model_id)
model = AutoModelForCausalLM.from_pretrained(peft_config.base_model_name_or_path)
model = PeftModel.from_pretrained(model, peft_model_id).to(device)
model.eval()


tokenizer = AutoTokenizer.from_pretrained(
    peft_config.base_model_name_or_path,
    add_bos_token=True
)


def uwaterloo_output(post_title, post_text):
    prompt = f"""
Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
            
### Instruction:
Respond to the reddit post in the style of a University of Waterloo student.

### Input:
{post_title}
{post_text}

### Response:
"""
    model_input = tokenizer(prompt, return_tensors="pt").to(device)
    with torch.no_grad():
        model_output = model.generate(**model_input, max_new_tokens=256, repetition_penalty=1.15)[0]
    output = tokenizer.decode(model_output, skip_special_tokens=True)
    return output.split('### Response:\n')[-1]


iface = gr.Interface(
    fn=uwaterloo_output,
    inputs=[
        gr.Textbox("", label="Post Title"),
        gr.Textbox("", label="Post Text"),
    ],
    outputs=gr.Textbox("", label="Mistraloo-SFT")
)

iface.launch()