stable-2023-08-29
Pre-release
Pre-release
kimishpatel
released this
29 Aug 16:58
·
5189 commits
to main
since this release
New models enabled (e2e tested via portable lib):
- Wav2Letter
- Inception V3 and Inception V4
- Resnet18 and Resnet50
Quantization:
- Enabled E2E MobileNet V2:
- Model can be quantized and run with portable + quantized op (for quantize/dequantize ops) lib.
Follow, https://github.com/pytorch/executorch/blob/main/examples/README.md#quantization, to run a quantized model via portable lib.
- Model can be quantized and run with portable + quantized op (for quantize/dequantize ops) lib.
- MobileNet V3:
- Needs bumping up the pytorch nightly version (dev20230828) in order to enable MobileNet V3 quantization. However, this breaks ViT export, hence this cut will skip MobileNet V3 quantization until we resolve ViT export breakage.
Delegation:
- API update:
- [breaking changes] delegate AOT APIs are moved from
executorch/backends/
toexecutorch/exir/backend
. To address the breakage: Updatefrom executorch.backends.backend_details
tofrom executorch.exir.backend.backend_details
, andfrom executorch.backends.backend_api
tofrom executorch.exir.backend.backend_api
- [breaking changes] delegate AOT APIs are moved from
- XNNPACK:
- XNNPACK delegated models can run on Mac/Linux in OSS
- XNNPACK lowering workflow examples have been added for MobileNet V2 (with quantization and delegation) and MobileNet V3 (with delegation)
- Showcase preliminary XNNPACK perf stats on Linux x86 & Mac M1
Selective build:
- Added buck2 examples to demonstrate 3 APIs to do selective build on any executorch runtime build
- Run test_selective_build.sh