app.py

import gradio as gr 
import os 
import json
import requests


HF_TOKEN = os.getenv('HF_TOKEN')
HEADERS = {"Authorization": f"Bearer {HF_TOKEN}"}

zephyr_7b_beta = os.getenv('zephyr_7b_beta')
zephyr_7b_alpha = os.getenv('zephyr_7b_alpha')


def build_input_prompt(message, chatbot):
    """
    Constructs the input prompt string from the chatbot interactions and the current message.
    """
    input_prompt = "<|system|>\n</s>\n<|user|>\n"
    for interaction in chatbot:
        input_prompt = input_prompt + str(interaction[0]) + "</s>\n<|assistant|>\n" + str(interaction[1]) + "\n</s>\n<|user|>\n"

    input_prompt = input_prompt + str(message) + "</s>\n<|assistant|>"
    return input_prompt


def post_request_beta(payload):
    """
    Sends a POST request to the predefined Zephyr-7b-Beta URL and returns the JSON response.
    """
    response = requests.post(zephyr_7b_beta, headers=HEADERS, json=payload)
    response.raise_for_status()  # Will raise an HTTPError if the HTTP request returned an unsuccessful status code
    return response.json()


def post_request_alpha(payload):
    """
    Sends a POST request to the predefined Zephyr-7b-Alpha URL and returns the JSON response.
    """
    response = requests.post(zephyr_7b_alpha, headers=HEADERS, json=payload)
    response.raise_for_status()  # Will raise an HTTPError if the HTTP request returned an unsuccessful status code
    return response.json()


def predict_beta(message, chatbot=[], temperature=0.9, max_new_tokens=256, top_p=0.6, repetition_penalty=1.0):
    temperature = float(temperature)
    top_p = float(top_p)

    input_prompt = build_input_prompt(message, chatbot)

    data = {
        "inputs": input_prompt,
        "parameters": {
            "max_new_tokens": max_new_tokens,
            "temperature": temperature,
            "top_p": top_p,
            "repetition_penalty": repetition_penalty,
            "do_sample": True,
        },
    }

    try:
        response_data = post_request_beta(data)
        json_obj = response_data[0]
        
        if 'generated_text' in json_obj and len(json_obj['generated_text']) > 0:
            bot_message = json_obj['generated_text']
            chatbot.append((message, bot_message))
            return "", chatbot
        elif 'error' in json_obj:
            raise gr.Error(json_obj['error'] + ' Please refresh and try again with smaller input prompt')
        else:
            warning_msg = f"Unexpected response: {json_obj}"
            raise gr.Error(warning_msg)
    except requests.HTTPError as e:
        error_msg = f"Request failed with status code {e.response.status_code}"
        raise gr.Error(error_msg)
    except json.JSONDecodeError as e:
        error_msg = f"Failed to decode response as JSON: {str(e)}"
        raise gr.Error(error_msg)


def predict_alpha(message, chatbot=[], temperature=0.9, max_new_tokens=256, top_p=0.6, repetition_penalty=1.0):
    temperature = float(temperature)
    top_p = float(top_p)

    input_prompt = build_input_prompt(message, chatbot)

    data = {
        "inputs": input_prompt,
        "parameters": {
            "max_new_tokens": max_new_tokens,
            "temperature": temperature,
            "top_p": top_p,
            "repetition_penalty": repetition_penalty,
            "do_sample": True,
        },
    }

    try:
        response_data = post_request_alpha(data)
        json_obj = response_data[0]
        
        if 'generated_text' in json_obj and len(json_obj['generated_text']) > 0:
            bot_message = json_obj['generated_text']
            chatbot.append((message, bot_message))
            return "", chatbot
        elif 'error' in json_obj:
            raise gr.Error(json_obj['error'] + ' Please refresh and try again with smaller input prompt')
        else:
            warning_msg = f"Unexpected response: {json_obj}"
            raise gr.Error(warning_msg)
    except requests.HTTPError as e:
        error_msg = f"Request failed with status code {e.response.status_code}"
        raise gr.Error(error_msg)
    except json.JSONDecodeError as e:
        error_msg = f"Failed to decode response as JSON: {str(e)}"
        raise gr.Error(error_msg)


def retry_fun_beta(chat_history_beta ):
    """
    Retries the prediction for the last message in the chat history.
    Removes the last interaction and gets a new prediction for the same message from Zephyr-7b-Beta
    """
    if not chat_history_beta or len(chat_history_beta) < 1:
        raise gr.Error("Chat history is empty or invalid.")
    
    message = chat_history_beta[-1][0]
    chat_history_beta.pop()
    _, updated_chat_history_beta = predict_beta(message, chat_history_beta)
    return updated_chat_history_beta


def retry_fun_alpha(chat_history_alpha ):
    """
    Retries the prediction for the last message in the chat history.
    Removes the last interaction and gets a new prediction for the same message from Zephyr-7b-Alpha
    """
    if not chat_history_alpha or len(chat_history_alpha) < 1:
        raise gr.Error("Chat history is empty or invalid.")
    
    message = chat_history_alpha[-1][0]
    chat_history_alpha.pop()
    _, updated_chat_history_alpha = predict_alpha(message, chat_history_alpha)
    return updated_chat_history_alpha


title = "🌀Zephyr Playground🎮"
description = """
Welcome to the Zephyr Playground! This interactive space lets you experience the prowess of two distinct Zephyr models – [Zephyr-7b-Alpha](https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha) and [Zephyr-7b-Beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) – side by side. These models are products of fine-tuning the Mistral models.

- 🔎 Dive deep into the nuances and performance of these models by comparing their responses in real-time.
- 📖 For a comprehensive understanding of the Zephyr models, delve into their [technical report](https://arxiv.org/abs/2310.16944) and experiment with the [official Zephyr demo](https://huggingfaceh4-zephyr-chat.hf.space/).
- 🛠 If you wish to explore more chat models or set up your own interactive demo, visit the [Hugging Face's chat playground](https://huggingface.co/spaces/HuggingFaceH4/chat-playground).
"""
footnote = """Note: All rights, including licensing and acceptable use policies, related to the Zephyr models, can be found on their respective model pages on Hugging Face.
"""

css = """
.gradio-container {
    width: 100vw !important;
    min-height: 100vh !important;
    padding:0 !important;
    margin:0 !important;
    max-width: none !important;
}
"""

# Create chatbot components
chat_beta = gr.Chatbot(label="zephyr-7b-beta", layout='panel')
chat_alpha = gr.Chatbot(label="zephyr-7b-alpha", layout='panel')

# Create input and button components
textbox = gr.Textbox(container=False,
                     placeholder='Enter text and click the Submit button or press Enter')
submit = gr.Button('Submit', variant='primary',)
retry = gr.Button('🔄Retry', variant='secondary')
undo = gr.Button('↩️Undo', variant='secondary')

# Layout the components using Gradio Blocks API
with gr.Blocks(css=css) as demo:
  gr.HTML(f'<h1><center> {title} </center></h1>')
  gr.Markdown(description)
  with gr.Row():
    chat_beta.render()
    chat_alpha.render()
  with gr.Group():
    with gr.Row(equal_height=True):
      with gr.Column(scale=5):
        textbox.render()
      with gr.Column(scale=1):
        submit.render()
    with gr.Row():
      retry.render()
      undo.render()
      clear = gr.ClearButton(value='🗑️Clear',
                            components=[textbox,
                                        chat_beta, 
                                        chat_alpha])

  gr.Markdown(footnote)

  # Assign events to components
  textbox.submit(predict_beta, [textbox, chat_beta], [textbox, chat_beta])
  textbox.submit(predict_alpha, [textbox, chat_alpha], [textbox, chat_alpha])
  submit.click(predict_beta, [textbox, chat_beta], [textbox, chat_beta])
  submit.click(predict_alpha, [textbox, chat_alpha], [textbox, chat_alpha])

  undo.click(lambda x:x[:-1], [chat_beta], [chat_beta])
  undo.click(lambda x:x[:-1], [chat_alpha], [chat_alpha])

  retry.click(retry_fun_beta, [chat_beta], [chat_beta])
  retry.click(retry_fun_alpha, [chat_alpha], [chat_alpha])

  gr.Examples([
              ['Hi! Who are you?'],
              ['What is a meme?'],
              ['Explain the plot of Cinderella in a sentence.'],
              ['Assuming I am a huge alien species with the ability to consume helicopters, how long would it take me to eat one?'],
              ],
              textbox)
              
  
# Launch the demo
demo.launch(debug=True)