Official implementation of "Searching for Winograd-aware Quantized Networks" (MLSys'20)

oneAPI Deep Neural Network Library (oneDNN)

ncnn is a high-performance neural network inference framework optimized for the mobile platform

Winograd minimal convolution algorithm generator for convolutional neural networks.

A compiler for lowering quantized ML operators to AMD AI Engine (AIE) firmware.

A heterogeneous accelerator-centric compute cluster

Firmware tools for Unitree Go2

Differentiable architecture search for convolutional and recurrent networks

Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models"

SAM: Sharpness-Aware Minimization (PyTorch)

A binary fault Injector based on TensorFI fault injector. For more details, please refer to our paper

CGRA-Flow is an integrated framework for CGRA compilation, exploration, synthesis, and development.

Open deep learning compiler stack for cpu, gpu and specialized accelerators

Latency performance measurement framework for ExecuTorch models with SME2 acceleration. Enables operator-level performance analysis, bottleneck identification, and automated reporting.

Generate versal system design from ONNX model. AI engine kernels. Sub-microsecond speeds for autoencoders.

Open-source AI acceleration on FPGA: from ONNX to RTL

Run Time for AIE and FPGA based platforms

Linux device tree generator for the Xilinx SDK (Vivado > 2014.1)

Mastering Embedded Linux Development Fourth Edition, published by Packt

CHARM: Composing Heterogeneous Accelerators on Heterogeneous SoC Architecture

MaxEVA: Maximizing the Efficiency of Matrix Multiplication on Versal AI Engine (accepted as full paper at FPT'23)

Linux-based partitioning hypervisor

A machine learning accelerator core designed for energy-efficient AI at the edge.

[NeurIPS 2020] MCUNet: Tiny Deep Learning on IoT Devices; [NeurIPS 2021] MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning; [NeurIPS 2022] MCUNetV3: On-Device Training Under 2…

Sleep detector using kria KV260 AI vision and Bluecoin

A tool to deploy Deep Neural Networks on PULP-based SoC's

[ICML 2025 Spotlight] Enhancing Certified Robustness via Block Reflector Orthogonal Layers and Logit Annealing Loss

Dataflow compiler for QNN inference on FPGAs