Strage answer in the GET result

{‘text’: ‘Create an Italian sentence of about 35 words in a extremely kind tone in which you ask federico to comeback, now’}
Create an Italian sentence of about 35 words in a extremely kind tone in which you ask federico to comeback, now
127.0.0.1 - - [10/Mar/2023 16:40:18] “POST / HTTP/1.1” 302 -
127.0.0.1 - - [10/Mar/2023 16:40:18] “GET /?result=+that+he+has+been+away%0A%0A%22Federico%2C+spero+che+tu+possa+tornare+presto+da+noi%2C+sarebbe+bello+vederti+di+nuovo+qui+e+condividere+insieme+un+po%27+di+allegria%21%22 HTTP/1.1” 200 -

I had this strange error. I sent the sentence ‘text’ but in the GET it added the first part in italian, that is strange and I don’t know where it’s taken it.

Is it a bug of the api or did I make some mistakes?

1 Like

Welcome to the community @ileniadangelo94

First things first. The completions endpoint take POST requests not GET.

The model finished your prompt with the most likely tokens until stop sequence was reached.

If you want it to translate right after the text ends, consider modifying your prompt. Appending "\nTranslation:" is a step in the right direction.

Read on prompt design in the docs

3 Likes

Hi! Thank you so much for your answer! Sorry I’m very new in this stuff, what do you mean with: “The completions endpoint take POST requests not GET.” I’m totally not an expert in these methods, sorry for the dumb question

No problem. In that case, I would recommend using playground and reading docs.

2 Likes