search

google / aqt

Apache License 2.0
262 stars 27 forks source link

Set the QTensor's dequant_dtype during SERVE mode to scale_t's dtype. #538

Closed copybara-service[bot] closed 7 months ago

copybara-service[bot] commented 7 months ago

Set the QTensor's dequant_dtype during SERVE mode to scale_t's dtype.