[Hexagon] F2qi avgpool bug fix

[rasagna-quic]

We see an accuracy issue when FQ2I is enabled for the avg_pool2d op in the deeplab_v3 QAT model. The error seems to be happening because the underlying graph has the output of the avg_pool mapped directly to the conv2d without a quantize-dequantize layer in between.

Graph of interest

graph TD;
W(dequantize: A_scale, A_zero_point) --> Conv2d;
A(dequantize: B_scale, B_zero_point) --> avg_pool2d;
avg_pool2d --> Conv2d;
Conv2d --> Output(quantize: C_scale, C_zero_point);

Loading

Quantization parameters after Fq2i pass:
qnn.avg_pool2d (Bs, Bz, Cs, Cz)
qnn.conv2d (Cs, Cz, As, Az)
qnn.requantize (Cs*As, 0, Cs, Cz)

Proposed fix for quantization parameters:
qnn.avg_pool2d (Bs, Bz, Bs, Bz)
qnn.conv2d (Bs, Bz, As, Az)
qnn.requantize (Cs*As, 0, Cs, Cz)

where quantized parameters As, Bs, Cs are scales and Az, Bz, Cz are zeropoints.

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• No users to auto-tag found, no teams are specified in PR title _{See #10317 for details}

_{Generated by tvm-bot}

[rasagna-quic]

cc: @quic-sanirudh @jverma-quic @kparzysz-quic

[rasagna-quic]

@tvm-bot rerun

[quic-sanirudh]

@tvm-bot rerun

[kparzysz-quic]

Thanks!