app.py

import spaces
import os
import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
from huggingface_hub import snapshot_download
import torch
from accelerate import Accelerator

# Initialize Accelerator for efficient multi-GPU/Zero optimization
accelerator = Accelerator()

# Load the model and tokenizer
model_path = snapshot_download(
    repo_id=os.environ.get("REPO_ID", "SimpleBerry/LLaMA-O1-Supervised-1129")
)

tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    torch_dtype=torch.float16,
    device_map="auto"
).eval()

DESCRIPTION = '''
# SimpleBerry/LLaMA-O1-Supervised-1129 | Optimized for Streaming and Hugging Face Zero Space.
This model is experimental and focused on advancing AI reasoning capabilities.

**To start a new chat**, click "clear" and begin a fresh dialogue.
'''

LICENSE = """
--- MIT License ---
"""

template = "<start_of_father_id>-1<end_of_father_id><start_of_local_id>0<end_of_local_id><start_of_thought><problem>{content}<end_of_thought><start_of_rating><positive_rating><end_of_rating>\n<start_of_father_id>0<end_of_father_id><start_of_local_id>1<end_of_local_id><start_of_thought><expansion>"

def llama_o1_template(data):
    text = template.format(content=data)
    return text


@spaces.GPU
def gen_one_token(inputs,temperature,top_p):
    output = model.generate(
            **inputs,
            max_new_tokens=1,
            temperature=temperature,
            top_p=top_p,
            do_sample=True,
            use_cache=True,
            pad_token_id=tokenizer.eos_token_id,
            return_dict_in_generate=True,
            output_scores=False
        )
    return output


def generate_text(message, history, max_tokens=512, temperature=0.9, top_p=0.95):
    input_text = llama_o1_template(message)
    for i in range(max_tokens):
        inputs = tokenizer(input_text, return_tensors="pt").to(accelerator.device)
        output = gen_one_token(inputs,temperature,top_p)
        # Return text with special tokens included
        generated_text = tokenizer.decode(output, skip_special_tokens=False)
        input_text += generated_text
        yield generated_text

with gr.Blocks() as demo:
    gr.Markdown(DESCRIPTION)

    chatbot = gr.ChatInterface(
        generate_text,
        title="SimpleBerry/LLaMA-O1-Supervised-1129 | Optimized Demo",
        description="Adjust settings below as needed.",
        examples=[
            ["How many r's are in the word strawberry?"],
            ['If Diana needs to bike 10 miles to reach home and she can bike at a speed of 3 mph for two hours before getting tired, and then at a speed of 1 mph until she reaches home, how long will it take her to get home?'],
            ['Find the least odd prime factor of $2019^8+1$.'],
        ],
        cache_examples=False,
        fill_height=True
    )

    with gr.Accordion("Adjust Parameters", open=False):
        max_tokens_slider = gr.Slider(minimum=128, maximum=2048, value=512, step=1, label="Max Tokens")
        temperature_slider = gr.Slider(minimum=0.1, maximum=1.5, value=0.9, step=0.1, label="Temperature")
        top_p_slider = gr.Slider(minimum=0.05, maximum=1.0, value=0.95, step=0.01, label="Top-p (nucleus sampling)")

    gr.Markdown(LICENSE)

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