Valid OpenApi Key Not Accepted with df.sektch.apply

[approximatelabs]

Valid OpenApi Key Not Accepted with df.sektch.apply

[bluecoconut]

Hmm, I'm not sure.

1 thing to note: your key is visible in the screenshot, so you will probably want to roll that key asap.

it looks like the env var is getting noticed... so my best guess is something about cutting the key a character short or something when copying? Or having a deactivated key?

you can probably quickly test this (without using sketch apply) by just using lambdaprompt directly

import lambdaprompt
import os
os.environ['OPENAI_API_KEY'] = '***'

lambdaprompt.GPT3Prompt("Test completion after this line:")()

it will probably give you the same error (with the same wrong key) and then you can get one that works?

I just ran the above example, with a fresh key, and it worked immediately.

[bluecoconut]

I think this isn't a bug with the library, and i just realized you could probably iterate with the api key directly on openai endpoint as well (just use requests and env vars, and see if your key works)

import requests
import os
os.environ["OPENAI_API_KEY"] = "***"
headers = {
        "Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}",
        "Content-Type": "application/json",
    }

data = {
    "prompt": "Hello...",
    "model": "text-davinci-003",
}

resp = requests.post("https://api.openai.com/v1/completions", headers=headers, json=data)
print(resp.text)

If this works for you, then setting up environment variables the same way with sketch doesn't -- then i can re-open this issue (be sure to reach out again here). For now i'm going to close the issue.