How much disk does each of the bloom models require?
#50
by
dgaff
- opened
Hey all - I'm looking for a listing of the new models by disk usage and I can't seem to find that - is there anywhere where I can find that?
A good rule of thumb for autoregressive transformers is 13x the number of parameters for training and 2x the number of parameters for inference.
It needed around 400GB just to fit the all the weights files. They list the sizes of the weights and checkpoints under the Training section.
Closing as this seem resolved.
TimeRobber
changed discussion status to
closed