jackwish commented on a change in pull request #4629: [QNN] Channel wise
quantization - Quantize & Requantize
URL: https://github.com/apache/incubator-tvm/pull/4629#discussion_r363578316
##########
File path: src/relay/qnn/op/quantize.cc
##########
@@ -71,13 +78,29 @@ Expr MakeQuantize(Expr data, Expr output_scale, Expr
output_zero_point, DataType
}
Expr QuantizeLower(const Expr& input_tensor, const Expr& output_scale,
- const Expr& output_zero_point, const QuantizeAttrs* attrs) {
+ const Expr& output_zero_point, const Array<IndexExpr>&
input_shape,
+ const QuantizeAttrs* attrs) {
const auto out_dtype = attrs->out_dtype;
+ const auto axis = attrs->axis;
+
+ size_t n_dim = input_shape.size();
+
+ auto expanded_output_scale = output_scale;
+ if (!IsConstScalar(output_scale)) {
+ expanded_output_scale = ExpandBiasToMatchAxis(output_scale, n_dim, {axis});
Review comment:
what about renaming `ExpandBiasToMatchAxis` to something more generic?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
