Comparison to https://github.com/sashabaranov/go-gpt3

[prologic]

Can you briefly explain and add to the README how your librayr compares to https://github.com/sashabaranov/go-gpt3 ?

Thanks!

[tylermann]

It's probably a good idea, I actually haven't looked at that one or used it as this one was created before it. It looks fairly similar, although doesn't support streaming completion responses from what I can tell. Maybe if someone else has experience with both they could submit a PR to add a section on this.

[prologic]

I've only used your library so far so that person is not me 😆

[dabdine]

Having used both, the two main things I can see that are missing are:

1. Embeddings
2. Moderations

Embeddings are a current issue for me so i'll likely push a PR to add that, which will bring closer parity to sashabaranov/go-gpt3

Edit: Also, the library is using all the deprecated endpoints. Newer endpoints don't use the /engines prefix any longer (engines itself is deprecated, now they have a models endpoint, and the model is sent along with the request)

[dabdine]

A PR for embeddings has been added:
#11

[yorinasub17]

I just wanted to chime in to mention that for me, I was originally using sashabaranov/go-gpt3, but switched to this library because the API errors returned were not structs which makes it hard to write code that bubble errors back to the user (e.g., if the user entered prompt uses too many tokens).