gpt3_macro
gpt3_macro copied to clipboard
Cache generated code to reduce API usage
Currently, on incremental builds, an Api request would be made. Having some sort of local cache would help prevent running into any API limits usage for the stated case. Something like cacache alongside the description (and maybe an optional source path) as the cache key could serve some good :)
And then maybe exposing an optional feature flag to invalidate/not utilize the cache during build wouldn't be bad also.
But where's the fun if it's not a different block of code every time with potentially different behavior?