from threading import Thread

import gradio as gr
import torch
from transformers import (AutoModelForCausalLM, AutoTokenizer,
                          TextIteratorStreamer)

MODEL_ID = "alibayram/gemma3-tr-v64k-it"

# Model ve tokenizer yükleme
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

def build_prompt(gecmis, kullanici_mesaji):
    mesajlar = []

    mesajlar.extend(gecmis)

    mesajlar.append({
        "role": "user",
        "content": kullanici_mesaji
    })

    return tokenizer.apply_chat_template(
        mesajlar,
        tokenize=False,
        add_generation_prompt=True,
    )


def respond(
    mesaj,
    gecmis: list[dict[str, str]],
    max_tokens,
    temperature,
    top_p,
):
    prompt = build_prompt(gecmis, mesaj)

    girisler = tokenizer(prompt, return_tensors="pt").to(model.device)

    streamer = TextIteratorStreamer(
        tokenizer,
        skip_prompt=True,
        skip_special_tokens=True,
    )

    uretim_parametreleri = dict(
        **girisler,
        streamer=streamer,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
    )

    thread = Thread(target=model.generate, kwargs=uretim_parametreleri)
    thread.start()

    cevap = ""
    for token in streamer:
        cevap += token
        yield cevap


chatbot = gr.ChatInterface(
    respond,
    type="messages",
    additional_inputs=[
        gr.Slider(1, 1024, value=64, step=1, label="Maksimum Yeni Token"),
        gr.Slider(0.1, 1.99, value=0.7, step=0.1, label="Sıcaklık (Temperature)"),
        gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p"),
    ],
)

with gr.Blocks() as demo:
    chatbot.render()

if __name__ == "__main__":
    demo.launch()