# AI software accelerator framework
Source: [https://docs.qualcomm.com/doc/80-63195-1/topic/AI-software-accelerator-framework.html](https://docs.qualcomm.com/doc/80-63195-1/topic/AI-software-accelerator-framework.html)
Qualcomm® Neural Processing SDK and AI Model Efficiency Toolkit
(AIMET) support hardware acceleration.
## Qualcomm® Neural Processing SDK
The [Qualcomm Neural Processing SDK](https://developer.qualcomm.com/software/qualcomm-neural-processing-sdk/tools)allow developers to quickly integrate AI/ML models
into their Android apps by abstracting hardware complexities, and providing the advantage of
fast, portable AI application development. Qualcomm® Neural Processing
SDK can be used for :
- Convert Caffe, Caffe2, [TensorFlow](https://tensorflow.org/), [PyTorch](https://pytorch.org/) and TensorFlow Lite models to a [Deep Learning Container (DLC)](https://docs.qualcomm.com/doc/80-63195-1/topic/Deep-Learning-Container--DLC-.html) file
- The [quantization](https://docs.qualcomm.com/doc/80-63195-1/topic/quantization.html) of DLC files to 8bit/16bit fixed point for
execution on the Qualcomm® Hexagon™ Tensor Processor (HTP)
- Integrate a network into Android apps via C++ or Java
- Execute the network on the Qualcomm® Kryo™ CPU, the Qualcomm® Adreno™ GPU, or the HTP
- Debug and analyze the performance of the ML model
Model training is performed on any popular deep learning framework that is supported by the
SDK. After training is complete, the trained model is converted into a [DLC](https://docs.qualcomm.com/doc/80-63195-1/topic/Deep-Learning-Container--DLC-.html) file that can be loaded into the SDK runtime that runs on the target
device.
In source code and tools, Qualcomm® Neural Processing SDK for AI is
referred to as Snapdragon Neural Processing Engine (SNPE).
Figure : Typical workflow for AI development
## AI Model Efficiency Toolkit (AIMET)
[AIMET](https://quic.github.io/aimet-pages/index.html) is a library that provides advanced model quantization and compression
techniques for trained neural network models. It provides features that have been proven to
improve run-time performance of deep learning, neural network models with lower compute and
memory requirements and minimal impact to task accuracy.
AIMET is designed to work with [PyTorch](https://pytorch.org/) and [TensorFlow](https://tensorflow.org/) models.
Qualcomm® Innovation Center also maintains the [AIMET Model Zoo](https://github.com/quic/aimet-model-zoo)
open-source repository which is a collection of popular neural network models optimized for
8-bit inference. Recipes for quantizing floating point models using AIMET are also provided.
For more details on AIMET, please visit: [https://github.com/quic/aimet](https://github.com/quic/aimet).
**Parent Topic:** [AI resources](https://docs.qualcomm.com/doc/80-63195-1/topic/AI-resources.html)
Last Published: May 16, 2024
[Previous Topic
AI hardware cores/accelerators](https://docs.qualcomm.com/bundle/publicresource/80-63195-1/topics/AI-hardware-cores-accelerators.md) [Next Topic
Qualcomm Neural Processing SDK for AI](https://docs.qualcomm.com/bundle/publicresource/80-63195-1/topics/Qualcomm-Neural-Processing-SDK-for-AI.md)