Skip to content

fix: quantize int8 in float32 to avoid float16 rounding and overflow - #406

Closed
selimfedakar wants to merge 1 commit into
MinishLab:mainfrom
selimfedakar:fix/int8-quantize-float32-precision
Closed

fix: quantize int8 in float32 to avoid float16 rounding and overflow#406
selimfedakar wants to merge 1 commit into
MinishLab:mainfrom
selimfedakar:fix/int8-quantize-float32-precision

Commits

  1. Commits on Nov 2, 2026