Trying to get chat completion API to work yet receiving a AttributeError: text, how do I fix this?

API
1

Hi I am trying to run my code that integrates with the chat completions API (OpenAI API) however when I try to submit a prompt to it it throws this error:

AttributeError: text

This error is specific to this part of the code :

Extract the generated response text from the API response

chat_response = response.choices[0].text.strip()

How do I fix this so that all of my code runs and I am able to generate chat responses?

Here is all of my code:
app = Flask(name)

Set your OpenAI API key here

openai.api_key = “sk-my-api-key”

Define the prompt and chat history

prompt = “prompt”
messages = [
{“role”: “system”, “content”: “What is the capital of France?”},
{“role”: “user”, “content”: “The capital of France is Paris.”},
{“role”: “assistant”, “content”: “The Los Angeles Dodgers won the World Series in 2020.”},
{“role”: “user”, “content”: “Where was it played?”}
]

Define a model and engine

model_engine = “gpt-3.5-turbo”

Generate the response using the OpenAI API

response = openai.ChatCompletion.create(
model=model_engine,
temperature=0.7,
max_tokens=1024,
n=1,
stop=None,
messages=messages,
)

Print the chatbot response

print(messages)

@app.route(“/”)
def index():
return render_template(“index.html”)

@app.route(“/chat”, methods=[“POST”])
def chat():
# Get the user’s prompt from the form data
prompt = request.form[“prompt”]

# Define the chat history
chat_history = [{'role': 'system', 'content': "What is the capital of France?"}, 
                {'role': 'user', 'content': "The capital of France is Paris."}, 
                {'role': 'assistant', 'content': "The Los Angeles Dodgers won the World Series in 2020."}, 
                {'role': 'user', 'content': "Where was it played?"}]

# Generate the response using the OpenAI API
response = openai.ChatCompletion.create(
    model="gpt-3.5-turbo",
    temperature=0.7,
    max_tokens=1024,
    n=1,
    stop=None,
    messages=messages,
)

# Extract the generated response text from the API response
chat_response = response.choices[0].text.strip()

# Render the chatbot response template with the response text
return render_template("chat.html", prompt=prompt, response=response)

if name == “main”:
app.run(debug=True)

2

The response format is different than for the standard text completions (see this example from the docs):

{
 'id': 'chatcmpl-6p9XYPYSTTRi0xEviKjjilqrWU2Ve',
 'object': 'chat.completion',
 'created': 1677649420,
 'model': 'gpt-3.5-turbo',
 'usage': {'prompt_tokens': 56, 'completion_tokens': 31, 'total_tokens': 87},
 'choices': [
   {
    'message': {
      'role': 'assistant',
      'content': 'The 2020 World Series was played in Arlington, Texas at the Globe Life Field, which was the new home stadium for the Texas Rangers.'},
    'finish_reason': 'stop',
    'index': 0
   }
  ]
}

So instead of

chat_response = response.choices[0].text.strip()

You need to get

chat_response = response.choices[0].message.content.strip()
3

Thank you, I’ll try is and let you know if it works

4

now my response is always the same:
{ “choices”: [ { “finish_reason”: “stop”, “index”: 0, “message”: { “content”: “The 2020 World Series was played in Arlington, Texas at Globe Life Field.”, “role”: “assistant” } } ], “created”: 1684060885, “id”: “chatcmpl-7G3SHBIWSvqBBoBlNSAtiKn0944fO”, “model”: “gpt-3.5-turbo-0301”, “object”: “chat.completion”, “usage”: { “completion_tokens”: 17, “prompt_tokens”: 55, “total_tokens”: 72 } }

What do I add for ID and created?

5

Ah, no no you don’t use this whole thing. It was just an example response.

Here is your code updated, try this new version, exactly like this:

# Define the chat history
chat_history = [{'role': 'system', 'content': "What is the capital of France?"}, 
                {'role': 'user', 'content': "The capital of France is Paris."}, 
                {'role': 'assistant', 'content': "The Los Angeles Dodgers won the World Series in 2020."}, 
                {'role': 'user', 'content': "Where was it played?"}]

# Generate the response using the OpenAI API
response = openai.ChatCompletion.create(
    model="gpt-3.5-turbo",
    temperature=0.7,
    max_tokens=1024,
    n=1,
    stop=None,
    messages=chat_history, # This also needed changing, as you weren't using your own history above
)

# Extract the generated response text from the API response
chat_response = response.choices[0].message.content.strip() # This is what's different from using non-chat completions

# Render the chatbot response template with the response text
return render_template("chat.html", prompt=prompt, response=response)
6

Its still giving me a response on the front end, that looks like this: Chatbot: { “choices”: [ { “finish_reason”: “stop”, “index”: 0, “message”: { “content”: “I’m sorry, I didn’t see your notes. Please provide me with your notes and I will be happy to create a presentation for you.”, “role”: “assistant” } } ], “created”: 1684066304, “id”: “chatcmpl-7G4rgMtktUBXSkgRw7kH2U7pcSSTP”, “model”: “gpt-3.5-turbo-0301”, “object”: “chat.completion”, “usage”: { “completion_tokens”: 29, “prompt_tokens”: 118, “total_tokens”: 147 } }

7

This is my code now:

app = Flask(__name__)

# Set your OpenAI API key here
openai.api_key = "sk-my-qpi-key"

# Define the prompt and chat history
prompt = "make a presentation from my notes"
chat_history = [
    {"role": "system", "content": "I take your notes and thematically organize them into a presentation, breaking the content down into individual slides."},
    {"role": "user", "content": "Please take these notes a make a multi-slide presentation for me."},
    {"role": "assistant", "content": "I write organized and easy to read presentations, that organize your slide notes into individual slides with headers and body conent."},
    {"role": "user", "content": "can you create a presentation for me, breaking my notes down into slides with headers and body content for each slide?"},
    {"role": "assistant", "content": "Of course please provide me with your notes."}
]

# Define a model and engine
model_engine = "gpt-3.5-turbo"

# Generate the response using the OpenAI API
response = openai.ChatCompletion.create(
    model=model_engine,
    temperature=0.7,
    max_tokens=1024,
    n=1,
    stop=None,
    messages=chat_history,
)

# Print the chatbot response
print(chat_history)

@app.route("/")
def index():
    return render_template("index.html")


@app.route("/chat", methods=["POST"])
def chat():
    # Get the user's prompt from the form data
    prompt = request.form["prompt"]

    # Define the chat history
    chat_history = [{"role": "system", "content": "I take your notes and thematically organize them into a presentation, breaking the content down into individual slides."},
                    {"role": "user", "content": "Please take these notes a make a multi-slide presentation for me."},
                    {"role": "assistant", "content": "I write organized and easy to read presentations, that organize your slide notes into individual slides with headers and body conent."},
                    {"role": "user", "content": "can you create a presentation for me, breaking my notes down into slides with headers and body content for each slide?"},
                    {"role": "assistant", "content": "Of course please provide me with your notes."}]

    # Generate the response using the OpenAI API
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        temperature=0.7,
        max_tokens=1024,
        n=1,
        stop=None,
        messages=chat_history,
    )

    # Extract the generated response text from the API response
    chat_response = response.choices[0].message.content.strip()

    # Render the chatbot response template with the response text
    return render_template("chat.html", prompt=prompt, response=response)

if __name__ == "__main__":
    app.run(debug=True)
8

Its generating a response, but i dont think it is generating a response from my prompt, only the chat history, I have a simply HTML page where i can enter a prompt, in this case presentation notes, and it does not seem to read the notes and make a presentation from them, intstead it makes a presenattion about a random topic. I would like it to make a presentaiton from my notes lol

9

Actually to be honest. I’ve been testing it. I don’t think it’s receiving my prompt at all. As each time I test it is responds to me based on exactly where we left off in our conversation in the chat_history. It is never able to generate a pre sooner related to any content I attempt to submit. Is there a reason for this? How do I fix it?

10

You’re not actually adding your prompt to the chat_history that you are sending to OpenAI. How is openAI supposed to be aware of your prompt this way?

image
image661×512 25.1 KB

You need to add your prompt as another user type message to the array, before you add it to the openai call.

Look, I think you need a proper programmer. This forum is not going to help you rewrite your entire program: the problems you’re highlighting have nothing to do with openAI, but with the rest of your code, so I would advise you to get help from a programmer to build your application.