app.py

import os
import time
from typing import List, Tuple, Optional, Dict

import google.generativeai as genai
import gradio as gr
from PIL import Image

print("google-generativeai:", genai.__version__)

GG_API_KEY = os.environ.get("GG_API_KEY")
oaiusr = os.environ.get("OAI_USR")
oaipwd = os.environ.get("OAI_PWD")

TITLE = """<h2 align="center">Tomoniai's Gemini Pro Chat</h2>"""

AVATAR_IMAGES = ("./user.png", "./botg.png")

IMAGE_WIDTH = 512


def preprocess_stop_sequences(stop_sequences: str) -> Optional[List[str]]:
    if not stop_sequences:
        return None
    return [sequence.strip() for sequence in stop_sequences.split(",")]


def preprocess_image(image: Image.Image) -> Optional[Image.Image]:
    image_height = int(image.height * IMAGE_WIDTH / image.width)
    return image.resize((IMAGE_WIDTH, image_height))


def preprocess_chat_history(
    history: List[Tuple[Optional[str], Optional[str]]]
) -> List[Dict[str, List[str]]]:
    messages = []
    for user_message, model_message in history:
        if user_message is not None:
            messages.append({'role': 'user', 'parts': [user_message]})
        if model_message is not None:
            messages.append({'role': 'model', 'parts': [model_message]})
    return messages


def user(text_prompt: str, chatbot: List[Tuple[str, str]]):
    return "", chatbot + [[text_prompt, None]]


def bot(
    google_key: str,
    image_prompt: Optional[Image.Image],
    temperature: float,
    max_output_tokens: int,
    stop_sequences: str,
    top_k: int,
    top_p: float,
    chatbot: List[Tuple[str, str]]
):
    google_key = google_key if google_key else GG_API_KEY

    text_prompt = chatbot[-1][0]
    genai.configure(api_key=google_key)
    generation_config = genai.types.GenerationConfig(
        temperature=temperature,
        max_output_tokens=max_output_tokens,
        stop_sequences=preprocess_stop_sequences(stop_sequences=stop_sequences),
        top_k=top_k,
        top_p=top_p)

    if image_prompt is None:
        model = genai.GenerativeModel('gemini-pro')
        response = model.generate_content(
            preprocess_chat_history(chatbot),
            stream=True,
            generation_config=generation_config)
        response.resolve()
    else:
        image_prompt = preprocess_image(image_prompt)
        model = genai.GenerativeModel('gemini-pro-vision')
        response = model.generate_content(
            contents=[text_prompt, image_prompt],
            stream=True,
            generation_config=generation_config)
        response.resolve()

    # streaming effect
    chatbot[-1][1] = ""
    for chunk in response:
        for i in range(0, len(chunk.text), 10):
            section = chunk.text[i:i + 10]
            chatbot[-1][1] += section
            time.sleep(0.01)
            yield chatbot


image_prompt_component = gr.Image(type="pil", label="Image", scale=1, height=400)
chatbot_component = gr.Chatbot(
    label='Gemini',
    bubble_full_width=False,
    avatar_images=AVATAR_IMAGES,
    scale=2,
    height=400
)
text_prompt_component = gr.Textbox(
    placeholder="Hi there!",
    label="Ask me anything and press Enter"
)
run_button_component = gr.Button()
temperature_component = gr.Slider(
    minimum=0,
    maximum=1.0,
    value=0.4,
    step=0.05,
    label="Temperature",
    info=(
        "Temperature controls the degree of randomness in token selection. Lower "
        "temperatures are good for prompts that expect a true or correct response, "
        "while higher temperatures can lead to more diverse or unexpected results. "
    ))
max_output_tokens_component = gr.Slider(
    minimum=1,
    maximum=2048,
    value=1024,
    step=1,
    label="Token limit",
    info=(
        "Token limit determines the maximum amount of text output from one prompt. A "
        "token is approximately four characters. The default value is 2048."
    ))
stop_sequences_component = gr.Textbox(
    label="Add stop sequence",
    value="",
    type="text",
    placeholder="STOP, END",
    info=(
        "A stop sequence is a series of characters (including spaces) that stops "
        "response generation if the model encounters it. The sequence is not included "
        "as part of the response. You can add up to five stop sequences."
    ))
top_k_component = gr.Slider(
    minimum=1,
    maximum=40,
    value=32,
    step=1,
    label="Top-K",
    info=(
        "Top-k changes how the model selects tokens for output. A top-k of 1 means the "
        "selected token is the most probable among all tokens in the model’s "
        "vocabulary (also called greedy decoding), while a top-k of 3 means that the "
        "next token is selected from among the 3 most probable tokens (using "
        "temperature)."
    ))
top_p_component = gr.Slider(
    minimum=0,
    maximum=1,
    value=1,
    step=0.01,
    label="Top-P",
    info=(
        "Top-p changes how the model selects tokens for output. Tokens are selected "
        "from most probable to least until the sum of their probabilities equals the "
        "top-p value. For example, if tokens A, B, and C have a probability of .3, .2, "
        "and .1 and the top-p value is .5, then the model will select either A or B as "
        "the next token (using temperature). "
    ))

user_inputs = [
    text_prompt_component,
    chatbot_component
]

bot_inputs = [
    image_prompt_component,
    temperature_component,
    max_output_tokens_component,
    stop_sequences_component,
    top_k_component,
    top_p_component,
    chatbot_component
]

with gr.Blocks() as demo:
    gr.HTML(TITLE)
    with gr.Column():
        with gr.Row():
            image_prompt_component.render()
            chatbot_component.render()
        text_prompt_component.render()
        run_button_component.render()
        with gr.Accordion("Parameters", open=False):
            temperature_component.render()
            max_output_tokens_component.render()
            stop_sequences_component.render()
            with gr.Accordion("Advanced", open=False):
                top_k_component.render()
                top_p_component.render()

    run_button_component.click(
        fn=user,
        inputs=user_inputs,
        outputs=[text_prompt_component, chatbot_component],
        queue=False
    ).then(
        fn=bot, inputs=bot_inputs, outputs=[chatbot_component],
    )

    text_prompt_component.submit(
        fn=user,
        inputs=user_inputs,
        outputs=[text_prompt_component, chatbot_component],
        queue=False
    ).then(
        fn=bot, inputs=bot_inputs, outputs=[chatbot_component],
    )

demo.queue(max_size=99).launch(auth=(oaiusr, oaipwd),show_api=False, debug=False, show_error=True)