Target group
- Data scientists in research
- Data scientists in industry
Your requirements
- You have a trained model for an application.
- You want to calculate a large amount of data for inference tasks in an energy-efficient and fast way.
- The model can be optimised and quantised.
Our offer
In cooperation with users of KISSKI, the suitability of a model for data processing
and its feasibility for FPGA-based systems is evaluated and, if necessary, implemented.Requirements
A trained model is necessary for the use of the FPGA computing resources, since it is a question of accelerating the inference.
Consultation regarding the compatibility of the neural network and the task is necessary.
The overall application in the form of a model and example/user data should already be worked out.
Subsequently, an optimisation and quantisation of the network is necessary.
