Cory Stephenson

@yar_vol @abhi_venigalla @MosaicML One of the authors here... here's the loss for the first 1000 iterations of a run we just did on 256 GPUs starting from init. Loss for diffusion models doesn't always correlate with sample quality, so we're doing the work to really prove things are working for an upcoming blog :)

@abhi_venigalla @MosaicML Did you try training on the largest scale? We actually want to see the loss drop and checkpoints, who knows maybe you forgot about interconnect (just kidding, but please launch the jobs and show the weights:)

We're coming for all the models! This week our Vision team profiled Stable Diffusion on @MosaicML Cloud and found that training from scratch costs <$160k, and can be done in under 2 weeks. mosaicml.com/blog/training-…