Go directly to the demo of a portable ML solution (object detection, TF, MobileNets, COCO)
Webcam-based live test
ML/SW/HW autotuning scoreboard
Our mission is to help researchers and practitioners find reproduced papers with portable, reusable and open-source components
(models, data sets, frameworks, libraries, tools) for emerging digital technologies (AI, ML, quantum, IoT),
assemble functional solutions
, understand how to use them,
and participate in collaborative benchmarking and optimization
of novel techniques using live scoreboards
Here you can find and try several real use cases.
Note that this is an ongoing community project and there is still a lot to be done -
please get in touch if you have suggestions or encounter any issues.
Participate in collaborative ML&systems benchmarking
Preparing, submitting and reproducing ML benchmarking results
is a very tedious, ad-hoc and time consuming process.
Check these CK solutions to learn how to automatically install and run real applications from the MLPerf benchmark on your platform in a few relatively simple steps:
You can then participate in collaborative benchmarking to validate MLPerf results
and submit the new ones using these live CK scoreboards
You can also see all dependencies
on reusable components from this portal required to assemble this portable solution.
Test the above object detection solution in practice
Besides looking at benchmarking results, we also want to test research techniques in practice on real data sets.
You can test how the above object detection solution works in practice
in your browser
See other reproduced results (AI, ML, quantum, IoT)
Follow this link
to find reproduced results from open competitions, reproducible hackathons and collaborative benchmarking efforts
we helped to organize
Check our concept of a live research paper
See the live paper
with reusable workflows to apply machine learning to compilers (our collaboration with the Raspberry Pi foundation).
Create your own dashboard for crowd-benchmarking
Please follow this documentaiton
about how to create your own customized dashboard for crowd-benchmarking and live research papers.
Create your own portable CK solution for crowd-benchmarking