3 ms·

I think openAI might be experimenting with smaller context lengths to save on costs or something since I've had a few other things break down like this for me t

I think openAI might be experimenting with smaller context lengths to save on costs or something since I've had a few other things break down like this for me today too (even in GPT-4)

api 3y ago
That’s what I thought too. I like using local models and those with short contexts will definitely go off into cuckooland if you start scrolling off the end of the buffer.