app.py

import sys, os
import gradio as gr

## if kgen not exist
try:
    import kgen
except:
    GH_TOKEN = os.getenv("GITHUB_TOKEN")
    git_url = f"https://{GH_TOKEN}@github.com/KohakuBlueleaf/TIPO-KGen@tipo"

    ## call pip install
    os.system(f"pip install git+{git_url}")

import re
import random
from time import time

import torch
import huggingface_hub
from transformers import set_seed

if sys.platform == "win32":
    # dev env in windows, @spaces.GPU will cause problem
    def GPU(func, *args, **kwargs):
        return func

else:
    from spaces import GPU

import kgen.models as models
import kgen.executor.tipo as tipo
from kgen.formatter import seperate_tags, apply_format
from kgen.generate import generate

from diff import load_model, encode_prompts
from meta import DEFAULT_NEGATIVE_PROMPT, DEFAULT_FORMAT


sdxl_pipe = load_model("OnomaAIResearch/Illustrious-xl-early-release-v0")

models.load_model(
    "KBlueLeaf/TIPO-500M-ft",
    device="cuda",
)
generate(max_new_tokens=4)
DEFAULT_TAGS = """
1girl, king halo (umamusume), umamusume,
ningen mame, ciloranko, ogipote, misu kasumi,
solo, leaning forward, sky,
masterpiece, absurdres, sensitive, newest
""".strip()
DEFAULT_NL = """
An illustration of a girl
""".strip()


def format_time(timing):
    total = timing["total"]
    generate_pass = timing["generate_pass"]

    result = ""

    result += f"""
### Process Time
| Total    | {total:5.2f} sec / {generate_pass:5} Passes | {generate_pass/total:7.2f} Passes Per Second|
|-|-|-|
"""
    if "generated_tokens" in timing:
        total_generated_tokens = timing["generated_tokens"]
        total_input_tokens = timing["input_tokens"]
    if "generated_tokens" in timing and "total_sampling" in timing:
        sampling_time = timing["total_sampling"] / 1000
        process_time = timing["prompt_process"] / 1000
        model_time = timing["total_eval"] / 1000

        result += f"""| Process  | {process_time:5.2f} sec / {total_input_tokens:5} Tokens | {total_input_tokens/process_time:7.2f} Tokens Per Second|
| Sampling | {sampling_time:5.2f} sec / {total_generated_tokens:5} Tokens | {total_generated_tokens/sampling_time:7.2f} Tokens Per Second|
| Eval     | {model_time:5.2f} sec / {total_generated_tokens:5} Tokens | {total_generated_tokens/model_time:7.2f} Tokens Per Second|
"""

    if "generated_tokens" in timing:
        result += f"""
### Processed Tokens:
* {total_input_tokens:} Input Tokens
* {total_generated_tokens:} Output Tokens
"""
    return result


@GPU(duration=10)
@torch.no_grad()
def generate(
    tags,
    nl_prompt,
    black_list,
    temp,
    output_format,
    target_length,
    top_p,
    min_p,
    top_k,
    seed,
    escape_brackets,
):
    default_format = DEFAULT_FORMAT[output_format]
    tipo.BAN_TAGS = [t.strip() for t in black_list.split(",") if t.strip()]
    generation_setting = {
        "seed": seed,
        "temperature": temp,
        "top_p": top_p,
        "min_p": min_p,
        "top_k": top_k,
    }
    inputs = seperate_tags(tags.split(","))
    if nl_prompt:
        if "<|extended|>" in default_format:
            inputs["extended"] = nl_prompt
        elif "<|generated|>" in default_format:
            inputs["generated"] = nl_prompt
    input_prompt = apply_format(inputs, default_format)
    if escape_brackets:
        input_prompt = re.sub(r"([()\[\]])", r"\\\1", input_prompt)

    meta, operations, general, nl_prompt = tipo.parse_tipo_request(
        seperate_tags(tags.split(",")),
        nl_prompt,
        tag_length_target=target_length,
        generate_extra_nl_prompt="<|generated|>" in default_format or not nl_prompt,
    )
    t0 = time()
    for result, timing in tipo.tipo_runner_generator(
        meta, operations, general, nl_prompt, **generation_setting
    ):
        result = apply_format(result, default_format)
        if escape_brackets:
            result = re.sub(r"([()\[\]])", r"\\\1", result)
        timing["total"] = time() - t0
        yield result, input_prompt, format_time(timing)


@GPU(duration=20)
@torch.no_grad()
def generate_image(
    seed,
    prompt,
    prompt2,
):
    torch.cuda.empty_cache()
    set_seed(seed)
    prompt_embeds, pooled_embeds2 = (
        encode_prompts(sdxl_pipe, prompt2, DEFAULT_NEGATIVE_PROMPT)
    )
    result2 = sdxl_pipe(
        prompt_embeds=prompt_embeds[0:1],
        negative_prompt_embeds=prompt_embeds[1:],
        pooled_prompt_embeds=pooled_embeds2[0:1],
        negative_pooled_prompt_embeds=pooled_embeds2[1:],
        num_inference_steps=24,
        width=1024,
        height=1024,
        guidance_scale=6.0,
    ).images[0]
    yield result2, None

    prompt_embeds,  pooled_embeds2 = (
        encode_prompts(sdxl_pipe, prompt, DEFAULT_NEGATIVE_PROMPT)
    )
    set_seed(seed)
    result = sdxl_pipe(
        prompt_embeds=prompt_embeds[0:1],
        negative_prompt_embeds=prompt_embeds[1:],
        pooled_prompt_embeds=pooled_embeds2[0:1],
        negative_pooled_prompt_embeds=pooled_embeds2[1:],
        num_inference_steps=24,
        width=1024,
        height=1024,
        guidance_scale=6.0,
    ).images[0]
    torch.cuda.empty_cache()
    yield result2, result


if __name__ == "__main__":
    with gr.Blocks(theme=gr.themes.Soft()) as demo:
        with gr.Accordion("Introduction and Instructions", open=False):
            gr.Markdown(
                """
## TIPO Demo
**The model for demo is 500M version with 4epoch training (25B token seen)**

### What is this
TIPO is a tool to extend, generate, refine the input prompt for T2I models.
<br>It can work on both Danbooru tags and Natural Language. Which means you can use it on almost all the existed T2I models.
<br>You can take it as "pro max" version of [DTG](https://huggingface.co/KBlueLeaf/DanTagGen-delta-rev2)

### How to use this demo
1. Enter your tags(optional): put the desired tags into "danboru tags" box
2. Enter your NL Prompt(optional): put the desired natural language prompt into "Natural Language Prompt" box
3. Enter your black list(optional): put the desired black list into "black list" box
4. Adjust the settings: length, temp, top_p, min_p, top_k, seed ...
4. Click "TIPO" button: you will see refined prompt on "result" box
5. If you like the result, click "Generate Image From Result" button
    * You will see 2 generated images, left one is based on your prompt, right one is based on refined prompt
    * The backend is diffusers, there are no weighting mechanism, so Escape Brackets is default to False

### Why inference code is private? When will it be open sourced?
1. This model/tool is still under development, currently is early Alpha version.
2. I'm doing some research and projects based on this.
3. The 200M model is released under CC-BY-NC-ND License currently. If you have interest, you can implement inference by yourself.
4. Once the project/research are done, I will open source all these models/codes with Apache2 license.

### Notification
**TIPO is NOT a T2I model. It is Prompt Gen, or, Text-to-Text model. 
<br>The generated images come from OnomaAIResearch/Illustrious-xl-early-release-v0 SDXL-based model**
"""
            )
        with gr.Row():
            with gr.Column(scale=5):
                with gr.Row():
                    with gr.Column(scale=3):
                        tags_input = gr.TextArea(
                            label="Danbooru Tags",
                            lines=7,
                            show_copy_button=True,
                            interactive=True,
                            value=DEFAULT_TAGS,
                            placeholder="Enter danbooru tags here",
                        )
                        nl_prompt_input = gr.Textbox(
                            label="Natural Language Prompt",
                            lines=7,
                            show_copy_button=True,
                            interactive=True,
                            value=DEFAULT_NL,
                            placeholder="Enter Natural Language Prompt here",
                        )
                        black_list = gr.TextArea(
                            label="Black List (seperated by comma)",
                            lines=4,
                            interactive=True,
                            value="monochrome",
                            placeholder="Enter tag/nl black list here",
                        )
                    with gr.Column(scale=2):
                        output_format = gr.Dropdown(
                            label="Output Format",
                            choices=list(DEFAULT_FORMAT.keys()),
                            value="Both, tag first (recommend)",
                        )
                        target_length = gr.Dropdown(
                            label="Target Length",
                            choices=["very_short", "short", "long", "very_long"],
                            value="long",
                        )
                        temp = gr.Slider(
                            label="Temp",
                            minimum=0.0,
                            maximum=1.5,
                            value=0.5,
                            step=0.05,
                        )
                        top_p = gr.Slider(
                            label="Top P",
                            minimum=0.0,
                            maximum=1.0,
                            value=0.95,
                            step=0.05,
                        )
                        min_p = gr.Slider(
                            label="Min P",
                            minimum=0.0,
                            maximum=0.2,
                            value=0.05,
                            step=0.01,
                        )
                        top_k = gr.Slider(
                            label="Top K", minimum=0, maximum=120, value=60, step=1
                        )
                        with gr.Row():
                            seed = gr.Number(
                                label="Seed",
                                minimum=0,
                                maximum=2147483647,
                                value=20090220,
                                step=1,
                            )
                            escape_brackets = gr.Checkbox(
                                label="Escape Brackets", value=False
                            )
                        submit = gr.Button("TIPO!", variant="primary")
                with gr.Accordion("Speed statstics", open=False):
                    cost_time = gr.Markdown()
            with gr.Column(scale=5):
                result = gr.TextArea(
                    label="Result", lines=8, show_copy_button=True, interactive=False
                )
                input_prompt = gr.Textbox(
                    label="Input Prompt", lines=1, interactive=False, visible=False
                )
                gen_img = gr.Button(
                    "Generate Image from Result", variant="primary", interactive=False
                )
                with gr.Row():
                    with gr.Column():
                        img1 = gr.Image(label="Original Prompt", interactive=False)
                    with gr.Column():
                        img2 = gr.Image(label="Generated Prompt", interactive=False)

        def generate_wrapper(*args):
            yield "", "", "", gr.update(interactive=False),
            for i in generate(*args):
                yield *i, gr.update(interactive=False)
            yield *i, gr.update(interactive=True)

        submit.click(
            generate_wrapper,
            [
                tags_input,
                nl_prompt_input,
                black_list,
                temp,
                output_format,
                target_length,
                top_p,
                min_p,
                top_k,
                seed,
                escape_brackets,
            ],
            [
                result,
                input_prompt,
                cost_time,
                gen_img,
            ],
            queue=True,
        )

        def generate_image_wrapper(seed, result, input_prompt):
            for img1, img2 in generate_image(seed, result, input_prompt):
                yield img1, img2, gr.update(interactive=False)
            yield img1, img2, gr.update(interactive=True)

        gen_img.click(
            generate_image_wrapper,
            [seed, result, input_prompt],
            [img1, img2, submit],
            queue=True,
        )
        gen_img.click(
            lambda *args: gr.update(interactive=False),
            None,
            [submit],
            queue=False,
        )

    demo.launch()