Max Ren

Reshape Conv2d Output

1

Reshape Convolution 2d output ```shell cd XNNPACK bash script/build-local.sh cd build/local make -j9 convolution-2d-test ./convolution-2d-test ```

Reshape Unpooling Output

1

Reshape Unpooling output

Reshape MaxPooling2d Output

2

Reshape max pool 2d output operator. Testing: ```shell cd XNNPACK bash scripts/build-local.sh cd build/local make -j9 max-pooling-2d-reshape-test ./make-pooling-2d-reshape-test ```

[PyTorch][ExportedProgram] add 'is_lifted_tensor_constant' and 'get_lifted_tensor_constant' utils

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #120546 as title Differential Revision: [D54149274](https://our.internmc.facebook.com/intern/diff/D54149274/)

Break flag bc

3

Summary: XNNPACK broke Backwards Compatibility by forcing pooling operations to reduce dims, and introducing a flag to allow these operation to keep the dims. This is backwards breaking because previously...

fix examples/xnnpack/README

1

Summary: Updating XNNPACK's README - Clean up directory structure by removing xnn_executor_runner - Add commands for running examples via cmake Differential Revision: D56524907

`xnn_weights_cache_provider` look_up doesn't work?

3

I've recently been experimenting with XNNPACK's weight cache to reduce load time by caching packed weights and also reduce memory pressure for repeated weights across the same kernels. I was...

XNN_FLAG_KEEP_DIM not backwards compatible

1

Commit: https://github.com/google/XNNPACK/commit/23f386b5d7eca18cb98659f84483cd8f20662251 Introduces the XNN_FLAG_KEEP_DIM flag which makes global_sum_pooling, global_average_pooling, and static mean no longer backwards compatible. That is calls previously to xnn_define_global_average_pooling will reduce dims, when previous to this...

update delegate for derived bias qspec

2

Summary: In the below we added quantizer annotation to the bias of GEMM operations. We update some of our delegate lowering logic to allow for this change Differential Revision: D56959071

make quantized model tests better

3

Summary: XNNPACK should be testing with random inputs and quantized models should be calibrated before lowering and testing. Right now there are some numerical issues being debugged and tracked by:...