Skip to content
Better HN
Top
New
Best
Ask
Show
Jobs
Search
⌘K
undefined | Better HN
0 points
ancientworldnow
1y ago
0 comments
Share
This was trained to be run at FP8 with no quality loss.
0 comments
default
newest
oldest
hislaziness
1y ago
The model description on huggingface says - Model size - 12.2B params, Tensor type - BF16. Is the Tensor type different from the training param size?
j
/
k
navigate · click thread line to collapse