[Torch] Support returning quantized weights and bias for BYOC use cases - #9135

Merged

masahi merged 9 commits into

apache:mainfrom

masahi:torch-quantize-weight

Sep 28, 2021

Commits on Sep 27, 2021

[Torch] refactored the way is bias quantization done
masahi
committed
5617524
support returning 8bit weight
masahi
committed
88c65d7
add test
masahi
committed
1ebf646
add doc
masahi
committed
3d3e050
pylint
masahi
committed
cc518c5

Commits on Sep 28, 2021

return_int8_weight -> keep_quantized_weight
masahi
committed
978e21f
fixed for dynamic linear case
masahi
committed
4a6f8c3
remove test function call
masahi
committed
59e5e88
simplifying
masahi
committed
2b5ef07