tensorflow2-deeplabv3plus_xception

The release provides tf2 deeplabv3plus_xception model W8A8 percentile per-channel encoding file

tensorflow2-deeplabv3plus_mbnv2

The release provides tf2 deeplabv3plus_mbnv2 model W8A8 percentile per-channel encoding file

AIMET Model Zoo June 2023 release (1.5.0)

Release summary

• Minor clean-up of code to minimize errors from static analysis (pylint)
• Added more acceptance tests for a few pytorch segmentation models and applied minor fixes to previous acceptance tests
• Config file and documentation updates for a pytorch MobileBert models
• Releasing next package release of aimet model zoo (installable wheel file binaries)
• Added code, documentation and artifacts the following new Tensorflow models: MobileNet-EdgeTPU, DeepLabV3Plus_Xception_TF2 and DeepLabV3Plus_MBNV2_TF2

Artifacts summary

AIMET model zoo python wheel file for PyTorch and Tensorflow models and evaluators

Package Installation

Please click here for instructions to install the pre-built AIMET model zoo python package.

tensorflow1_mobilenetedgetpu_w8a8

FP32 ckpt zipped model: includes the .mate, .index and .data-00000-of-00001 files.

torch_mobilebert

PyTorch MobileBERT W8A8 and W4A8:

GLUE dataset:
SQUAD dataset

Torch Uniformer Classification

FP32 weights
W8A8 weights and encodings

Resnet50 w8a16

Release artifacts for torch Resnet 50 w8a16

MMAction2 BMN Action Recognition

This release contains artifacts of mmaction2 models

AIMET Model Zoo May 2023 release (1.4.0)

Release summary

• Added acceptance tests framework and test suite for most pytorch and some tf models
• Updates to PR CI/CD CMake and Jenkins files
• Added W8A16 quantization and artifacts for pytorch ResNet50 model
• Releasing next package release of aimet model zoo (installable wheel file binaries)
• Added code, documentation and artifacts the following new Pytorch models : UniFormer, MMAction2 BMN

Artifacts summary

AIMET model zoo python wheel file for PyTorch and TensorFlow models and evaluators

Package Installation

Please click here for instructions to install the pre-built AIMET model zoo python package.

AIMET Model Zoo April 2023 release (1.3.0)

Release summary

• Further cleaned up code to minimize errors from static analysis (pylint)
• CI/CD jenkins pipeline refactored to incorporate proper code quality checks
• Model refactoring and code updates for one pytorch model (SalsaNext)
• Releasing third package release of aimet model zoo (installable wheel file binaries)
• Added artifacts the following new PyTorch models: SalsaNext w4a8
• Added code, documentation and artifacts the following new Pytorch models : SegNet, EfficientNet-B0
• Added code, documentation and artifacts the following new TensorFlow models: MobileNetV2, ResNet50

Artifacts summary

AIMET model zoo python wheel file for PyTorch and TensorFlow models and evaluators

Package Installation

Please click here for instructions to install the pre-built AIMET model zoo python package.