pull down to refresh

100 sats \ 3 replies \ @SimpleStacker 4 Sep \ on: Stacker Saloon
OpenAI's servers must be really congested. My work has slowed to a crawl, since my main bottleneck right now is feeding millions of prompts into their API and getting the response.
That's why I'm posting so much on SN.... basically blocked at work
write
preview
30 sats \ 2 replies \ @k00b 4 Sep
they might be rate-limiting you?
reply
50 sats \ 1 reply \ @SimpleStacker 4 Sep
Possibly, but I seem well within my account's rate limits, and in the past when i've exceeded a limit I got a message from the API saying so. I've also done much greater throughput before without problems. Starting this Tuesday (after labor day weekend), many of my batch jobs have been sitting with no progress and timing out after 24 hours.
reply
21 sats \ 0 replies \ @SimpleStacker 4 Sep
I've resorted to using concurrent (not batch) requests for now, even though it's twice as expensive, because at least I can reliably get output. It's also a lot slower because the batch requests are processed in parallel
reply