import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
import torch
from threading import Thread

MODEL_ID = "uncensoredai/UncensoredLM-DeepSeek-R1-Distill-Qwen-14B"

# โหลดโมเดล
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    torch_dtype=torch.float16,
    device_map="auto"
)

def generate_text(prompt, temperature=0.8, top_p=0.9, max_new_tokens=512):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True)
    generation_kwargs = dict(
        **inputs,
        streamer=streamer,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
        max_new_tokens=max_new_tokens
    )
    thread = Thread(target=model.generate, kwargs=generation_kwargs)
    thread.start()
    output = ""
    for new_text in streamer:
        output += new_text
        yield output

with gr.Blocks(title="Uncensored DeepSeek Qwen 14B") as demo:
    gr.Markdown("## 🧠 Uncensored DeepSeek Qwen 14B")
    gr.Markdown("Thai & English Chatbot – Powered by Qwen 14B Distilled Model")

    with gr.Row():
        with gr.Column(scale=3):
            prompt = gr.Textbox(label="Input", placeholder="พิมพ์ข้อความที่นี่...", lines=3)
            temperature = gr.Slider(0.1, 1.5, value=0.8, step=0.1, label="Temperature")
            top_p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top P")
            max_new_tokens = gr.Slider(64, 2048, value=512, step=64, label="Max New Tokens")
            btn = gr.Button("Generate")

        with gr.Column(scale=5):
            output = gr.Textbox(label="AI Response", lines=20)

    btn.click(generate_text, inputs=[prompt, temperature, top_p, max_new_tokens], outputs=output)

demo.queue().launch()