app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import torch
from gradio_client import utils as client_utils

# پچ کردن تابع get_type
def patched_get_type(schema):
    if isinstance(schema, bool):
        return "any" if schema else "never"
    if "const" in schema:
        return repr(schema["const"])
    if "type" in schema:
        return schema["type"]
    return "any"

client_utils.get_type = patched_get_type

# چاپ نسخه‌های پکیج‌ها برای عیب‌یابی
print(f"Gradio version: {gr.__version__}")
print(f"Transformers version: {transformers.__version__}")
print(f"Torch version: {torch.__version__}")
print(f"Huggingface_hub version: {huggingface_hub.__version__}")

# Load model and tokenizer with force_download
model_name = "arshiaafshani/Arsh-llm"
tokenizer = AutoTokenizer.from_pretrained(model_name, force_download=True, resume_download=False)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, force_download=True, resume_download=False)

# تنظیم توکن‌های خاص
tokenizer.bos_token = "<sos>"
tokenizer.eos_token = "<|endoftext|>"

# Create pipeline
pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    device=0 if torch.cuda.is_available() else -1
)

def respond(message, chat_history, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
    chat_history = chat_history or []
    messages = [{"role": "system", "content": system_message}] + \
               [{"role": "user", "content": msg} for msg, _ in chat_history] + \
               [{"role": "user", "content": message}, {"role": "assistant", "content": ""}]
    prompt = tokenizer.apply_chat_template(messages, tokenize=False)
    
    output = pipe(
        prompt,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        repetition_penalty=repeat_penalty,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id
    )
    
    response = output[0]['generated_text'][len(prompt):].strip()
    chat_history.append((message, response))
    return chat_history

with gr.Blocks() as demo:
    gr.Markdown("# Arsh-LLM Demo")
    with gr.Row():
        with gr.Column():
            system_msg = gr.Textbox("You are Arsh, a helpful assistant by Arshia Afshani. You should answer the user carefully.", 
                                  label="System Message")
            max_tokens = gr.Slider(1, 4096, value=2048, step=1, label="Max Tokens")
            temperature = gr.Slider(0.1, 4.0, value=0.7, step=0.1, label="Temperature")
            top_p = gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p")
            top_k = gr.Slider(0, 100, value=40, step=1, label="Top-k")
            repeat_penalty = gr.Slider(0.0, 2.0, value=1.1, step=0.1, label="Repetition Penalty")

    chatbot = gr.Chatbot(height=500)
    msg = gr.Textbox(label="Your Message")
    clear = gr.Button("Clear")

    def submit_message(message, chat_history, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
        chat_history = chat_history or []
        response = respond(message, chat_history, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty)
        return "", response

    msg.submit(
        submit_message,
        [msg, chatbot, system_msg, max_tokens, temperature, top_p, top_k, repeat_penalty],
        [msg, chatbot]
    )
    clear.click(lambda: None, None, chatbot, queue=False)

if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)