Profile avatar
lukeingalls.bsky.social
Engineering manager @ kapwing.com
5 posts 29 followers 16 following
comment in response to post
Why not do queueing instead and just run jobs until you hit a rate limit? That ought to maximize usage whereas exponential backoff might not
comment in response to post
I’ve done this on snippets up to about 1000 lines with encouraging results so far. Even processing I’m chunks maintains good results with gpt4
comment in response to post
It feels like it’s all of the worst of old-Twitter all in one place.