Collective Knowledge experimental workflow to crowd-benchmark and crowd-tune Tiny DNN. Currently on stall due to lack of time and resources. Further info about collaborative DNN optimization:
C++
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.cm
docker
program
.ckr.json
.gitignore
CHANGES
README.md

README.md

Collective Knowledge repository for evaluating and optimising Tiny DNN

logo License

We need to have easily customizable Tiny DNN builds via JSON API to be able to plug it to our framework for collaborative benchmarking and optimization of workloads across diverse inputs and hardware provided by volunteers (see cKnowledge.org/ai, live repo and papers 1, 2, 3 for more details).

Note that at the moment we are very limited with resources and mainly focus on Caffe crowd-benchmarking and crowd-tuning.

Prerequisites

  • Collective Knowledge framework (@GitHub)

Installation

Obtain CK repository for Tiny DNN:

 $ ck pull repo:ck-tiny-dnn

Compile simple bench (on Linux/MacOS/Windows; Android G++ has some issues):

 $ ck compile program:tiny-dnn-bench --speed

Note, that it will automatically install TinyDNN package via CK. You can check that it was installed properly via

 $ ck show env --tags=tiny-dnn

Run this bench

 $ ck run program:tiny-dnn-bench