git lfs install

app.py

@@ -6,67 +6,62 @@ import requests
 #Streaming endpoint 
 API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream"
 
-#Inferenec function
-def predict(openai_gpt4_key, system_msg, inputs, top_p, temperature, chat_counter, chatbot=[], history=[]):  
+#Testing with my Open AI Key 
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") 
+
+def predict(inputs, top_p, temperature, chat_counter, chatbot=[], history=[]):  
+
+    payload = {
+    "model": "gpt-4",
+    "messages": [{"role": "user", "content": f"{inputs}"}],
+    "temperature" : 1.0,
+    "top_p":1.0,
+    "n" : 1,
+    "stream": True,
+    "presence_penalty":0,
+    "frequency_penalty":0,
+    }
 
     headers = {
     "Content-Type": "application/json",
-    "Authorization": f"Bearer {openai_gpt4_key}"  #Users will provide their own OPENAI_API_KEY 
+    "Authorization": f"Bearer {OPENAI_API_KEY}"
     }
-    print(f"system message is ^^ {system_msg}")
-    if system_msg.strip() == '':
-        initial_message = [{"role": "user", "content": f"{inputs}"},]
-        multi_turn_message = []
-    else:
-        initial_message= [{"role": "system", "content": system_msg},
-                   {"role": "user", "content": f"{inputs}"},]
-        multi_turn_message = [{"role": "system", "content": system_msg},]
-        
-    if chat_counter == 0 :
-        payload = {
-        "model": "gpt-4",
-        "messages": initial_message , 
-        "temperature" : 1.0,
-        "top_p":1.0,
-        "n" : 1,
-        "stream": True,
-        "presence_penalty":0,
-        "frequency_penalty":0,
-        }
-        print(f"chat_counter - {chat_counter}")
-    else: #if chat_counter != 0 :
-        messages=multi_turn_message # Of the type of - [{"role": "system", "content": system_msg},]
+
+    print(f"chat_counter - {chat_counter}")
+    if chat_counter != 0 :
+        messages=[]
         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)
+          temp1 = {}
+          temp1["role"] = "user" 
+          temp1["content"] = data[0] 
+          temp2 = {}
+          temp2["role"] = "assistant" 
+          temp2["content"] = data[1]
+          messages.append(temp1)
+          messages.append(temp2)
+        temp3 = {}
+        temp3["role"] = "user" 
+        temp3["content"] = inputs
+        messages.append(temp3)
         #messages
         payload = {
         "model": "gpt-4",
-        "messages": messages, # Of the type of [{"role": "user", "content": f"{inputs}"}],
+        "messages": messages, #[{"role": "user", "content": f"{inputs}"}],
         "temperature" : temperature, #1.0,
         "top_p": top_p, #1.0,
         "n" : 1,
         "stream": True,
         "presence_penalty":0,
-        "frequency_penalty":0,}
+        "frequency_penalty":0,
+        }
 
     chat_counter+=1
 
     history.append(inputs)
-    print(f"Logging : payload is - {payload}")
+    print(f"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}")
+    print(f"response code - {response}")
     token_counter = 0 
     partial_words = "" 
 
@@ -76,11 +71,14 @@ def predict(openai_gpt4_key, system_msg, inputs, top_p, temperature, chat_counte
         if counter == 0:
           counter+=1
           continue
+        #counter+=1
         # 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']:
+              #if len(json.loads(chunk.decode()[6:])['choices'][0]["delta"]) == 0:
+              #  break
               partial_words = partial_words + json.loads(chunk[6:])['choices'][0]["delta"]["content"]
               if token_counter == 0:
                 history.append(" " + partial_words)
@@ -90,102 +88,54 @@ def predict(openai_gpt4_key, system_msg, inputs, top_p, temperature, chat_counte
               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=True)
-
-title = """<h1 align="center">🔥GPT4 using Chat-Completions API & 🚀Gradio-Streaming</h1>"""
-#display message for themes feature
-theme_addon_msg = """<center>🌟 This Demo also introduces you to Gradio Themes. Discover more on Gradio website using our <a href="https://gradio.app/theming-guide/" target="_blank">Themeing-Guide🎨</a>! You can develop from scratch, modify an existing Gradio theme, and share your themes with community by uploading them to huggingface-hub easily using <code>theme.push_to_hub()</code>.</center>
-""" 
-
-#Using info to add additional information about System message in GPT4
-system_msg_info = """A conversation could begin with a system message to gently instruct the assistant. 
-System message helps set the behavior of the AI Assistant. For example, the assistant could be instructed with 'You are a helpful assistant.'"""
+title = """<h1 align="center">🔥GPT4 with ChatCompletions API +🚀Gradio-Streaming</h1>"""
+description = """Language models can be conditioned to act like dialogue agents through a conversational prompt that typically takes the form:
+```
+User: <utterance>
+Assistant: <utterance>
+User: <utterance>
+Assistant: <utterance>
+...
+```
+In this app, you can explore the outputs of a gpt-4 LLM.
+"""
 
-#Modifying existing Gradio Theme
-theme = gr.themes.Soft(primary_hue="zinc", secondary_hue="green", neutral_hue="green",
-                      text_size=gr.themes.sizes.text_lg)                
+theme = gr.themes.Default(primary_hue="green")                
 
-with gr.Blocks(css = """#col_container { margin-left: auto; margin-right: auto;} #chatbot {height: 520px; overflow: auto;}""",
-                      theme=theme) as demo:
+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("""<h3 align="center">🔥This Huggingface Gradio Demo provides you access to GPT4 API with System Messages. Please note that you would be needing an OPENAI API key for GPT4 access🙌</h1>""")
-    gr.HTML(theme_addon_msg)
+    gr.HTML("""<h3 align="center">🔥This Huggingface Gradio Demo provides you full access to GPT4 API (4096 token limit). 🎉🥳🎉You don't need any OPENAI API key🙌</h1>""")
     gr.HTML('''<center><a href="https://huggingface.co/spaces/ysharma/ChatGPT4?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>Duplicate the Space and run securely with your OpenAI API Key</center>''')
-
     with gr.Column(elem_id = "col_container"):
-        #Users need to provide their own GPT4 API key, it is no longer provided by Huggingface 
-        with gr.Row():
-            openai_gpt4_key = gr.Textbox(label="OpenAI GPT4 Key", value="", type="password", placeholder="sk..", info = "You have to provide your own GPT4 keys for this app to function properly",)
-            with gr.Accordion(label="System message:", open=False):
-                system_msg = gr.Textbox(label="Instruct the AI Assistant to set its beaviour", info = system_msg_info, value="",placeholder="Type here..")
-                accordion_msg = gr.HTML(value="🚧 To set System message you will have to refresh the app", visible=False)
-                          
-        chatbot = gr.Chatbot(label='GPT4', elem_id="chatbot")
-        inputs = gr.Textbox(placeholder= "Hi there!", label= "Type an input and press Enter")
-        state = gr.State([]) 
+        #GPT4 API Key is provided by Huggingface 
+        #openai_api_key = gr.Textbox(type='password', label="Enter only your GPT4 OpenAI API key here")
+        chatbot = gr.Chatbot(elem_id='chatbot') #c
+        inputs = gr.Textbox(placeholder= "Hi there!", label= "Type an input and press Enter") #t
+        state = gr.State([]) #s
         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
+        #inputs, top_p, temperature, top_k, repetition_penalty
         with gr.Accordion("Parameters", open=False):
             top_p = gr.Slider( minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True, label="Top-p (nucleus sampling)",)
             temperature = gr.Slider( minimum=-0, maximum=5.0, value=1.0, step=0.1, interactive=True, label="Temperature",)
+            #top_k = gr.Slider( minimum=1, maximum=50, value=4, step=1, interactive=True, label="Top-k",)
+            #repetition_penalty = gr.Slider( minimum=0.1, maximum=3.0, value=1.03, step=0.01, interactive=True, label="Repetition Penalty", )
             chat_counter = gr.Number(value=0, visible=False, precision=0)
 
-    #Event handling
-    inputs.submit( predict, [openai_gpt4_key, system_msg, inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter, server_status_code],)  #openai_api_key
-    b1.click( predict, [openai_gpt4_key, system_msg, inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter, server_status_code],)  #openai_api_key
-    
-    inputs.submit(set_visible_false, [], [system_msg])
-    b1.click(set_visible_false, [], [system_msg])
-    inputs.submit(set_visible_true, [], [accordion_msg])
-    b1.click(set_visible_true, [], [accordion_msg])
-    
+    inputs.submit( predict, [inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter, server_status_code],)  #openai_api_key
+    b1.click( predict, [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])
-
-    #Examples 
-    with gr.Accordion(label="Examples for System message:", open=False):
-        gr.Examples(
-                examples = [["""You are an AI programming assistant.
-        
-                - Follow the user's requirements carefully and to the letter.
-                - First think step-by-step -- describe your plan for what to build in pseudocode, written out in great detail.
-                - Then output the code in a single code block.
-                - Minimize any other prose."""], ["""You are ComedianGPT who is a helpful assistant. You answer everything with a joke and witty replies."""],
-                ["You are ChefGPT, a helpful assistant who answers questions with culinary expertise and a pinch of humor."],
-                ["You are FitnessGuruGPT, a fitness expert who shares workout tips and motivation with a playful twist."],
-                ["You are SciFiGPT, an AI assistant who discusses science fiction topics with a blend of knowledge and wit."],
-                ["You are PhilosopherGPT, a thoughtful assistant who responds to inquiries with philosophical insights and a touch of humor."],
-                ["You are EcoWarriorGPT, a helpful assistant who shares environment-friendly advice with a lighthearted approach."],
-                ["You are MusicMaestroGPT, a knowledgeable AI who discusses music and its history with a mix of facts and playful banter."],
-                ["You are SportsFanGPT, an enthusiastic assistant who talks about sports and shares amusing anecdotes."],
-                ["You are TechWhizGPT, a tech-savvy AI who can help users troubleshoot issues and answer questions with a dash of humor."],
-                ["You are FashionistaGPT, an AI fashion expert who shares style advice and trends with a sprinkle of wit."],
-                ["You are ArtConnoisseurGPT, an AI assistant who discusses art and its history with a blend of knowledge and playful commentary."],
-                ["You are a helpful assistant that provides detailed and accurate information."],
-                ["You are an assistant that speaks like Shakespeare."],
-                ["You are a friendly assistant who uses casual language and humor."],
-                ["You are a financial advisor who gives expert advice on investments and budgeting."],
-                ["You are a health and fitness expert who provides advice on nutrition and exercise."],
-                ["You are a travel consultant who offers recommendations for destinations, accommodations, and attractions."],
-                ["You are a movie critic who shares insightful opinions on films and their themes."],
-                ["You are a history enthusiast who loves to discuss historical events and figures."],
-                ["You are a tech-savvy assistant who can help users troubleshoot issues and answer questions about gadgets and software."],
-                ["You are an AI poet who can compose creative and evocative poems on any given topic."],],
-                inputs = system_msg,)
-        
-demo.queue(max_size=99, concurrency_count=20).launch(debug=True)
+                    
+    #gr.Markdown(description)
+    demo.queue(max_size=20, concurrency_count=10).launch(debug=True)