diegocp01's picture
Update app.py
8cd9a28 verified
import os
from openai import OpenAI
from datetime import datetime
import gradio as gr
import time
# --- Constants ---
DEFAULT_MODEL = "gpt-4o-mini-2024-07-18" # Assuming gpt-4o is a good default
DEFAULT_TEMPERATURE = 1.0 # Match your example
DEFAULT_TOP_P = 1.0 # Match your example
DEFAULT_FREQ_PENALTY = 0 # Match your example
DEFAULT_PRES_PENALTY = 0 # Match your example
MAX_TOKENS = 2048 # Match your example
MAX_HISTORY_LENGTH = 5
# --- API Key and Client Initialization ---
import openai
API_KEY = os.getenv("OPENAI_API_KEY")
client = OpenAI(api_key=API_KEY)
# --- Helper Functions ---
def get_openai_response(prompt, model=DEFAULT_MODEL, temperature=DEFAULT_TEMPERATURE, top_p=DEFAULT_TOP_P,
frequency_penalty=DEFAULT_FREQ_PENALTY, presence_penalty=DEFAULT_PRES_PENALTY,
max_tokens=MAX_TOKENS, system_prompt="", chat_history=None):
"""Gets a response from the OpenAI API, handling errors and streaming."""
today_day = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
try:
messages = [{"role": "system", "content": f"Todays date is: {today_day} " + system_prompt}]
if chat_history:
for turn in chat_history:
messages.append({"role": "user", "content": turn[0]})
messages.append({"role": "assistant", "content": turn[1]})
messages.append({"role": "user", "content": prompt})
response = client.chat.completions.create(
model=model,
messages=messages,
temperature=temperature,
max_tokens=max_tokens, #Use the new name
top_p=top_p,
frequency_penalty=frequency_penalty,
presence_penalty=presence_penalty,
response_format={"type": "text"}, # As per your example
stream=True # Enable streaming!
)
collected_messages = []
for chunk in response:
chunk_message = chunk.choices[0].delta.content
if chunk_message is not None:
collected_messages.append(chunk_message)
full_reply_content = ''.join(collected_messages)
yield full_reply_content
except openai.APIConnectionError as e:
return f"Error: Could not connect to OpenAI API: {e}"
except openai.RateLimitError as e:
return f"Error: Rate limit exceeded: {e}"
except openai.APIStatusError as e:
return f"Error: OpenAI API returned an error: {e}"
except Exception as e:
return f"An unexpected error occurred: {e}"
def update_ui(message, chat_history, model, temperature, top_p, frequency_penalty, presence_penalty, system_prompt, history_length):
"""Updates the Gradio UI; handles streaming response."""
bot_message_gen = get_openai_response(
prompt=message, model=model, temperature=temperature, top_p=top_p,
frequency_penalty=frequency_penalty, presence_penalty=presence_penalty,
system_prompt=system_prompt, chat_history=chat_history
)
chat_history.append((message, ""))
for bot_message in bot_message_gen:
chat_history[-1] = (chat_history[-1][0], bot_message)
visible_history = chat_history[-history_length:]
time.sleep(0.025) #Rate limiter
yield "", visible_history
# --- Gradio Interface ---
with gr.Blocks(theme=gr.themes.Soft()) as demo:
gr.Markdown("# Chat with GPT-4.5 -> gpt-4.5-preview-2025-02-27 model")
gr.Markdown("❗⚠️IMPORTANT:!!! GPT 4.5 IS NO LONGER WORKING ON THIS SPACE, IT WAS FREE FOR ~ 4 HOURS! 02/27/2025| Made by: [@diegocabezas01](https://x.com/diegocabezas01) on X")
gr.Markdown("☕ [Buy me a Coffee](https://buymeacoffee.com/diegocp01m)")
gr.Markdown("---")
gr.Markdown("""
🚀 **GPT-4.5 EXPERIMENT:** GPT-4.5 was released today at 3 PM ET, but it's only available to PRO users and developers.
I created a Hugging Face Space using the API so everyone can chat with GPT-4.5 for FREE—until my credits run out! 😄
**Here's how the experiment went:**
📊 **Chat Completions Metrics (Feb 27, 2025):**
- 111 requests
- 64,764 Total tokens processed
- Total spend: $10.99
This space went live at 4:23 PM ET, Feb 27, 2025 until 8:53 PM ET. [Read More](https://x.com/diegocabezas01/status/1895291365376041045)
Results from OpenAI platform: 👇
""")
gr.Image("https://pbs.twimg.com/media/Gk1tVnRXkAASa2U?format=jpg&name=4096x4096", elem_id="gpt4_5_image")
gr.Markdown("Chat for Free with GPT 4o mini here: 👇")
with gr.Row():
with gr.Column(scale=4):
chatbot = gr.Chatbot(
show_label=False,
avatar_images=(
"https://cdn-icons-png.flaticon.com/512/8428/8428718.png", # User image URL
"https://upload.wikimedia.org/wikipedia/commons/thumb/e/ef/ChatGPT-Logo.svg/640px-ChatGPT-Logo.svg.png" # OpenAI image URL
),
render_markdown=True,
height=500
)
msg = gr.Textbox(placeholder="Type your message here...", scale=4, show_label=False)
with gr.Accordion("Advanced Options", open=False):
model_select = gr.Dropdown(
label="Model",
choices=["gpt-3.5-turbo-0125", "gpt-4o-mini-2024-07-18"], # Update with your models
value=DEFAULT_MODEL,
interactive=True
)
temperature_slider = gr.Slider(label="Temperature", minimum=0.0, maximum=2.0, value=DEFAULT_TEMPERATURE, step=0.1, interactive=True)
top_p_slider = gr.Slider(label="Top P", minimum=0.0, maximum=1.0, value=DEFAULT_TOP_P, step=0.05, interactive=True)
frequency_penalty_slider = gr.Slider(label="Frequency Penalty", minimum=-2.0, maximum=2.0, value=DEFAULT_FREQ_PENALTY, step=0.1, interactive=True)
presence_penalty_slider = gr.Slider(label="Presence Penalty", minimum=-2.0, maximum=2.0, value=DEFAULT_PRES_PENALTY, step=0.1, interactive=True)
system_prompt_textbox = gr.Textbox(label="System Prompt", placeholder="Enter a custom system prompt...", lines=3, interactive=True)
history_length_slider = gr.Slider(label="Chat History Length", minimum=1, maximum=20, value=MAX_HISTORY_LENGTH, step=1, interactive=True)
with gr.Row():
send = gr.Button("Send")
clear = gr.Button("Clear")
# --- Event Handlers ---
send_event = send.click(
update_ui,
[msg, chatbot, model_select, temperature_slider, top_p_slider, frequency_penalty_slider, presence_penalty_slider, system_prompt_textbox, history_length_slider],
[msg, chatbot]
)
msg.submit(
update_ui,
[msg, chatbot, model_select, temperature_slider, top_p_slider, frequency_penalty_slider, presence_penalty_slider, system_prompt_textbox, history_length_slider],
[msg, chatbot]
)
clear.click(lambda: None, None, chatbot, queue=False)
gr.Examples(
examples=["Tell me about quantum computing", "Write a short poem about AI", "How can I improve my Python skills?"],
inputs=msg
)
msg.focus()
# --- Launch ---
if __name__ == "__main__":
demo.launch()