Let a transformer block initialize its residual projections - #83
Merged
jessegrabowski merged 2 commits intoAug 11, 2026
Merged
Annotations
2 warnings
|
Run conda-incubator/setup-miniconda@8ee1f361103df19b6f8c8655fd3967a8ecb162d5
warning libmamba You are using 'pip' as an additional package manager.
Be aware that packages installed with 'pip' are managed independently from 'conda-forge' channel.
|
|
Run conda-incubator/setup-miniconda@8ee1f361103df19b6f8c8655fd3967a8ecb162d5
'nodefaults' channel detected: will remove 'defaults' if added implicitly. In the future, 'nodefaults' to remove 'defaults' won't be supported. Please set 'conda-remove-defaults' = 'true' in setup-miniconda to remove this warning.
|
background
wait
wait-all
cancel
parallel
Loading