Update app.py

app.py

@@ -1,105 +1,137 @@
 import gradio as gr
+import os 
+import json 
 import requests
-import json
 
-API_URL = "https://api.openai.com/v1/chat/completions"
 
-def predict(inputs, top_p, temperature, openai_api_key, chat_counter, chatbot=[], history=[]):
-    messages = format_messages(chatbot, inputs, chat_counter)
-    payload = create_payload(messages, top_p, temperature)
-    response = make_request(API_URL, openai_api_key, payload)
+#Streaming endpoint 
+API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream"
 
-    # Processar a resposta
-    new_chatbot, new_history, token_counter = process_response(response, history)
+#Testing with my Open AI Key 
+#OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") 
 
-    # Atualizar o chatbot e o histórico apenas se houver novos tokens
-    if token_counter > 0:
-        chatbot.extend(new_chatbot)
-        history = new_history
+def predict(inputs, top_p, temperature, openai_api_key, chat_counter, chatbot=[], history=[]):  #repetition_penalty, top_k
 
-    return chatbot, history, chat_counter
+    payload = {
+    "model": "gpt-4-1106-preview",
+    "messages": [{"role": "user", "content": f"{inputs}"}],
+    "temperature" : 1.0,
+    "top_p":1.0,
+    "n" : 1,
+    "stream": True,
+    "presence_penalty":0,
+    "frequency_penalty":0,
+    }
 
-def format_messages(chatbot, inputs, chat_counter):
-    messages = []
-    if chat_counter != 0:
-        for i in range(len(chatbot)):
-            user_message = {"role": "user", "content": chatbot[i][0]}
-            assistant_message = {"role": "assistant", "content": chatbot[i][1]}
-            messages.extend([user_message, assistant_message])
-    messages.append({"role": "user", "content": inputs})
-    return messages
+    headers = {
+    "Content-Type": "application/json",
+    "Authorization": f"Bearer {openai_api_key}"
+    }
 
-def create_payload(messages, top_p, temperature):
-    return {
+    print(f"chat_counter - {chat_counter}")
+    if chat_counter != 0 :
+        messages=[]
+        for data in chatbot:
+          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-4-1106-preview",
-        "messages": messages,
-        "temperature": temperature,
-        "top_p": top_p,
-        "n": 1,
+        "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,
-    }
+        "presence_penalty":0,
+        "frequency_penalty":0,
+        }
 
-def make_request(url, api_key, payload):
-    headers = {
-        "Content-Type": "application/json",
-        "Authorization": f"Bearer {api_key}"
-    }
-    response = requests.post(url, headers=headers, json=payload, stream=True)
-    return response
+    chat_counter+=1
 
-def process_response(response, history):
+    history.append(inputs)
+    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)
+    #response = requests.post(API_URL, headers=headers, json=payload, stream=True)
     token_counter = 0 
     partial_words = "" 
-    for chunk in response.iter_lines():
-        if chunk:
-            chunk_str = chunk.decode('utf-8').lstrip('data: ')
-            if chunk_str.strip() in ["[DONE]", "}"]:  # Adiciona } à verificação
-                break
-            try:
-                chunk_json = json.loads(chunk_str)
-                if 'choices' in chunk_json and len(chunk_json['choices']) > 0:
-                    chunk_data = chunk_json['choices'][0].get('delta', {})
-                    if 'content' in chunk_data:
-                        content = chunk_data['content']
-                        partial_words += content
-                        if token_counter == 0:
-                            history.append(" " + partial_words)
-                        else:
-                            history[-1] = partial_words
-                        token_counter += 1
-            except json.JSONDecodeError as e:
-                print("Error decoding JSON response:", e)
-                print("Raw chunk:", chunk_str)
-
-    chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2)]
-    return chat, history, token_counter
-
-
-
-def setup_ui():
-    with gr.Blocks() as demo:
-        with gr.Column():
-            openai_api_key = gr.Textbox(type='password', label="Insira sua chave de API OpenAI aqui")
-            chatbot = gr.Chatbot() 
-            inputs = gr.Textbox(placeholder="Olá!", label="Digite uma entrada e pressione Enter", lines=3)
-            state = gr.State([])
-            b1 = gr.Button(value="Executar", variant="primary")
-            
-            top_p = gr.Slider(minimum=0, maximum=1.0, value=1.0, step=0.05, label="Top-p")
-            temperature = gr.Slider(minimum=0, maximum=1.0, value=1.0, step=0.05, label="Temperature")
-            chat_counter = gr.Number(value=0, visible=False)
-
-        inputs.submit(predict, [inputs, top_p, temperature, openai_api_key, chat_counter, chatbot, state], [chatbot, state, chat_counter])
-        b1.click(predict, [inputs, top_p, temperature, openai_api_key, chat_counter, chatbot, state], [chatbot, state, chat_counter])
-
-    return demo
-
-def main():
-    demo = setup_ui()
-    demo.launch()
-
-if __name__ == "__main__":
-    main()
 
+    counter=0
+    for chunk in response.iter_lines():
+        #Skipping first chunk
+        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)
+              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  # resembles {chatbot: chat, state: history}  
+                   
+
+def reset_textbox():
+    return gr.update(value='')
+
+title = """<h1 align="center">🔥ChatGPT-4 Turbo API 🚀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-3.5-turbo LLM.
+"""
+
+css = """
+#col_container {width: 1000px; margin-left: auto; margin-right: auto;}
+#chatbot {height: 520px; overflow: auto;}
+"""
+
+with gr.Blocks(css=css) as demo:
+    gr.HTML(title)
+    #gr.HTML('''<center><a href="https://huggingface.co/spaces/ysharma/ChatGPTwithAPI?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"):
+        openai_api_key = gr.Textbox(type='password', label="Insira sua chave de API OpenAI aqui")
+        chatbot = gr.Chatbot(elem_id="chatbot") 
+        inputs = gr.Textbox(placeholder="Olá!", label="Digite uma entrada e pressione Enter", lines=3)
+        state = gr.State([]) 
+        b1 = gr.Button(value="Executar", variant="primary")
+    
+        #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)
+
+    inputs.submit( predict, [inputs, top_p, temperature, openai_api_key, chat_counter, chatbot, state], [chatbot, state, chat_counter],)
+    b1.click( predict, [inputs, top_p, temperature, openai_api_key, chat_counter, chatbot, state], [chatbot, state, chat_counter],)
+    b1.click(reset_textbox, [], [inputs])
+    inputs.submit(reset_textbox, [], [inputs])
+                    
+    #gr.Markdown(description)
+    demo.queue().launch(debug=True)