Hacker News new | ask | show | jobs
by mchiang 316 days ago
OpenAI has only provided MXFP4 weights. These are the same weights used by other cloud providers.
1 comments
irthomasthomas 316 days ago
Oh, I didn't know that. Weird!
link
reissbaker 316 days ago
It was natively trained in FP4. Probably both to reduce VRAM usage at inference time (fits on a single H100), and to allow better utilization of B200s (which are especially fast for FP4).
link
irthomasthomas 316 days ago
Interesting, thanks. I didn't know you could even train at FP4 on H100s
link
reissbaker 314 days ago
It's impressive they got it to work — the lowest I'd heard of this far was native FP8 training.
link