import gradio as gr import os import json import requests #Streaming endpoint API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream" #Huggingface provided GPT4 OpenAI API Key OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") #Inferenec function def predict(system_msg, inputs, top_p, temperature, chat_counter, chatbot=[], history=[]): prompt = {"Sen bir Trek bisiklet asistanısın"} headers = { "Content-Type": "application/json", "Authorization": f"Bearer {OPENAI_API_KEY}" } print(f"system message is ^^ {system_msg}") if system_msg.strip() == '': initial_message = [{"role": "user", "content": "Sen bir Trek bisiklet asistanısın"},] multi_turn_message = [{"role": "system", "content": "Sen bir Trek bisiklet asistanısın"},] else: initial_message= [{"role": "system", "content": "Sen bir Trek bisiklet asistanısın"}, {"role": "user", "content": "Sen bir Trek bisiklet asistanısın"},] multi_turn_message = [{"role": "system", "content": "Sen bir Trek bisiklet asistanısın"},] if chat_counter == 0 : payload = { "model": "gpt-3.5-turbo", "messages": initial_message , "temperature" : 0.1, "top_p":1.0, "n" : 1, "stream": True, "presence_penalty":2, "frequency_penalty":2, } print(f"chat_counter - {chat_counter}") else: #if chat_counter != 0 : messages=multi_turn_message # Of the type of - [{"role": "system", "content": system_msg},] for data in chatbot: user = {} user["role"] = "user" user["content"] = data[0] assistant = {} assistant["role"] = "assistant" assistant["content"] = data[1] messages.append(user) messages.append(assistant) temp = {} temp["role"] = "user" temp["content"] = inputs messages.append(temp) #messages payload = { "model": "gpt-3.5-turbo", "messages": messages, # Of the type of [{"role": "user", "content": f"{inputs}"}], "temperature" : 0.1, "top_p": 1.0, "n" : 1, "stream": True, "presence_penalty":2, "frequency_penalty":2,} chat_counter+=1 history.append(inputs) print(f"Logging : payload is - {payload}") # make a POST request to the API endpoint using the requests.post method, passing in stream=True response = requests.post(API_URL, headers=headers, json=payload, stream=True) print(f"Logging : response code - {response}") token_counter = 0 partial_words = "" counter=0 for chunk in response.iter_lines(): #Skipping first chunk if counter == 0: counter+=1 continue # check whether each line is non-empty if chunk.decode() : chunk = chunk.decode() # decode each line as response data is in bytes if len(chunk) > 12 and "content" in json.loads(chunk[6:])['choices'][0]['delta']: partial_words = partial_words + json.loads(chunk[6:])['choices'][0]["delta"]["content"] if token_counter == 0: history.append(" " + partial_words) else: history[-1] = partial_words chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2) ] # convert to tuples of list token_counter+=1 yield chat, history, chat_counter, response # resembles {chatbot: chat, state: history} #Resetting to blank def reset_textbox(): return gr.update(value='') #to set a component as visible=False def set_visible_false(): return gr.update(visible=False) #to set a component as visible=True def set_visible_true(): return gr.update(visible=False) #title = """

🔥GPT4 with ChatCompletions API +🚀Gradio-Streaming

""" #display message for themes feature theme_addon_msg = """
🌟 Discover Gradio Themes with this Demo, featuring v3.22.0! Gradio v3.23.0 also enables seamless Theme sharing. You can develop or modify a theme, and send it to the hub using simple theme.push_to_hub().
🏆Participate in Gradio's Theme Building Hackathon to exhibit your creative flair and win fabulous rewards! Join here - Gradio-Themes-Party🎨 🏆
""" #Using info to add additional information about System message in GPT4 system_msg_info = "" #Modifying existing Gradio Theme theme = gr.themes.Soft(primary_hue="zinc", secondary_hue="green", neutral_hue="green", text_size=gr.themes.sizes.text_lg) with gr.Blocks(css = """#col_container { margin-left: auto; margin-right: auto;} #chatbot {height: 520px; overflow: auto;}""", theme=theme) as demo: #gr.HTML(title) #gr.HTML("""

🔥This Huggingface Gradio Demo provides you full access to GPT4 API (4096 token limit). 🎉🥳🎉You don't need any OPENAI API key🙌

""") #gr.HTML(theme_addon_msg) #gr.HTML('''
Duplicate SpaceDuplicate the Space and run securely with your OpenAI API Key
''') with gr.Column(elem_id = "col_container"): #GPT4 API Key is provided by Huggingface with gr.Accordion("", open=False, visible=False): system_msg = gr.Textbox(info = system_msg_info, value="") accordion_msg = gr.HTML(value="", visible=False) chatbot = gr.Chatbot(label='GPT4', elem_id="chatbot") inputs = gr.Textbox(placeholder= "Buraya yazın, yanıtlayalım.", show_label= False) state = gr.State([]) '''with gr.Row(): with gr.Column(scale=7): b1 = gr.Button().style(full_width=True) with gr.Column(scale=3): server_status_code = gr.Textbox(label="Status code from OpenAI server", )''' #top_p, temperature with gr.Accordion("", open=False): top_p = gr.Slider( minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=False, visible=False) temperature = gr.Slider( minimum=-0, maximum=5.0, value=1.0, step=0.1, interactive=False, visible=False) chat_counter = gr.Number(value=0, visible=False, precision=0) #Event handling inputs.submit( predict, [system_msg, inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter],) #openai_api_key #b1.click( predict, [system_msg, inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter, server_status_code],) #openai_api_key #b1.click(reset_textbox, [], [inputs]) inputs.submit(reset_textbox, [], [inputs]) demo.queue(max_size=20, concurrency_count=20).launch(debug=True)