API Key

[LilianaCaG]

Hi everybody,
I noticed that Open AI just accept credit cards as payment method, I don't have one, but somehow I got an API Test Key, does it work for the Workshop? I also didn't get the confirmation E-Mail from Hugging Faces neither in my mailbox nor in my Spam folder, does it sound familiar to you? Apart from that, I think I got all installed. I hope that works.
Thank you and have a nice day!
Liliana

[TimBMK]

Hey Liliana,

You can check if the Hugging Face Registration was succesful by logging in on the page and retrieving your API key as described in the Setup. If you got an API key, everything should work as expected!

Regarding the OpenAI API key, I am not sure if the Test Key works, as this is the first time I'm hearing of this. To test if the key works, please run the following code (provided you installed the packages from the setup), with "APIKEY" replaced with your test key in "" (or loaded from the .env file):

from openai import OpenAI

client = OpenAI(api_key="APIKEY") # client = communicates with API

instructions = """ 
You are a sentiment classification system. 

You will be given a sentence and you have to determine if the sentence is "positive", "neutral", or "negative".

Only reply with "positive", "neutral", or "negative".
""" # the triple quotes allow us to have a multi-line string

messages = [
    # we pass the instructions as system message
    {
      "role": "system",
      "content": instructions
    },
    # and the content we wish to classify as user message
    {
      "role": "user",
      "content": "I love this movie!"
    }
]

response = client.chat.completions.create( # completions = create responses
    model="gpt-4o",
    messages=messages
)

response

If your key works, you will get a response from the OpenAI API. If it does not, the client should throw an error.

Unfortunately, I am not aware of a workaround for getting OpenAI API access without a credit card. Worst case, you might have to sit this one out - it will only be one exercise, and we'll go on to locally run LLMs afterwards.