Commit 218df759 authored by A. Unique TensorFlower's avatar A. Unique TensorFlower Committed by TensorFlower Gardener
Browse files

Added support for dequantization of inputs for some operators. 

When an operator's input is floating-point but the weights or bias are quantized, Dequantize operators are added to convert quantized inputs to floating-point since NNAPI does not support mixing float and quantized types. This is needed for models using post-training quantization.

Only CONV_2D and FULLY_CONNECTED operations support this for now.

PiperOrigin-RevId: 232699006
parent 1a43c97f
Show whitespace changes
Inline Side-by-side
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please to comment