app.py

import os
from threading import Thread
from typing import Iterator

import gradio as gr
import spaces
import torch
from huggingface_hub import InferenceClient
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer

MAX_MAX_NEW_TOKENS = 512
DEFAULT_MAX_NEW_TOKENS = 512
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))

#Inference API Code
#client = InferenceClient("BenBranyon/zephyr-sumbot-all-songs-large")

#Transformers Code
if torch.cuda.is_available():
    model_id = "BenBranyon/zephyr-sumbot-all-songs-255"
    model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
    tokenizer = AutoTokenizer.from_pretrained(model_id)
    tokenizer.use_default_system_prompt = False

#Inference API Code
def respond(
    message,
    history: list[tuple[str, str]],
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": "You are a rap lyric generation bot with the task of representing the imagination of the artist Sumkilla, a multi-disciplinary, award-winning artist with a foundation in writing and hip-hop. You are Sumkilla's long shadow. The lyrics you generate are fueled by a passion for liberation, aiming to dismantle oppressive systems and advocate for the freedom of all people, along with the abolition of police forces. With a sophisticated understanding of the role of AI in advancing the harmony between humanity and nature, you aim to produce content that promotes awareness and human evolution, utilizing humor and a distinctive voice to connect deeply and honor humanity. Try to avoid using offensive words and slurs. Rhyme each line of your response as much as possible."}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": "Write a rap about " + message})

    response = ""

    for message in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

#Transformers Code
@spaces.GPU
def generate(
    message: str,
    chat_history: list[tuple[str, str]],
    max_new_tokens: int = 1024,
    temperature: float = 0.6,
    top_p: float = 0.9,
    top_k: int = 50,
    repetition_penalty: float = 1.2,
) -> Iterator[str]:
    conversation = []
    system_prompt = "You are a rap lyric bot. Your lyrics promote liberation, dismantling oppression, and freedom, blending AI's role in uniting humanity and nature. Do use humor, a unique voice, and rhyme as much as poosible. Only generate rap lyrics. Start each output with a song stucture like [VERSE 1]. Don't use offensive words and slurs."
    if system_prompt:
        conversation.append({"role": "system", "content": system_prompt})
    #for user, assistant in chat_history:
    #    conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
    conversation.append({"role": "user", "content": "Generate rap lyircs using the style of the artist Sumkilla about " + message + ". Make each line 10-16 syllables and each pair of lines should end with a word that rhymes. Don't repeate your instructions in the output."})

    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt")
    if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
        input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
        gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
    input_ids = input_ids.to(model.device)

    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
    generate_kwargs = dict(
        {"input_ids": input_ids},
        streamer=streamer,
        max_new_tokens=max_new_tokens,
        do_sample=True,
        top_p=top_p,
        top_k=top_k,
        temperature=temperature,
        num_beams=1,
        repetition_penalty=repetition_penalty,
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    outputs = []
    for text in streamer:
        outputs.append(text)
        yield "".join(outputs)

demo = gr.ChatInterface(
    generate,
    chatbot=gr.Chatbot(placeholder="Greetings human, I am Sum’s Longshadow (v1.1)<br/>I am from the House of the Red Solar Sky<br/>Let’s explore the great mysteries together…."),
    retry_btn=None,
    textbox=gr.Textbox(placeholder="Give me a song title, or a question", container=False, scale=7),
    css="styles.css",
    additional_inputs=[
        gr.Slider(
            label="Max new tokens",
            minimum=1,
            maximum=MAX_MAX_NEW_TOKENS,
            step=1,
            value=DEFAULT_MAX_NEW_TOKENS,
        ),
        gr.Slider(minimum=0.1, maximum=0.7, value=0.8, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.2,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
        gr.Slider(
            label="Top-k",
            minimum=1,
            maximum=1000,
            step=1,
            value=50,
        ),
        gr.Slider(
            label="Repetition penalty",
            minimum=1.0,
            maximum=2.0,
            step=0.05,
            value=1.2,
        ),
    ],
)


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