I've spent the last couple days building out an automated classifier on top of the batch API, and just this morning (about a minute before the outage began!) started running my first live tests. I thought I was going mad!
8 comments
I've spent the last couple days building out an automated classifier on top of the batch API, and just this morning (about a minute before the outage began!) started running my first live tests. I thought I was going mad!
Welp, time to turn our brains on again ;)
How is this news when this happens almost every day?
Well if you look at the status page [0], it only seems to have become a daily occurrence in the past two weeks. Even before that uptime wasn't so good though.
You have no idea how glad I am to see you all here. xD
I asked it to add debug logs. Asked it to check debug logs. Stuck, locked me out.
Checked debug file, it was 1.6GB (!!!)
For a while I thought I was kicked out of the platform for violation of some T&C.
I don't understand how this is possible at all at Anthropic. Couldn't they, like, embed an agentic swarm into their backend that prevents any errors from ever making it into production? What am I missing?
As far as I heard from my people in there, Amodei does not believe in SREs.
Which, I suppose, figures.