You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some of the modules have custom quantization math embedded with old order of operations code, which is not consistent with PyTorch. Now that quant_refactor has changed the order of operations for FMS quantizers, these custom quantizers should call util functions in linear_utils to avoid bifurcation.
The text was updated successfully, but these errors were encountered:
Some of the modules have custom quantization math embedded with old order of operations code, which is not consistent with PyTorch. Now that quant_refactor has changed the order of operations for FMS quantizers, these custom quantizers should call util functions in linear_utils to avoid bifurcation.
The text was updated successfully, but these errors were encountered: