Julien Vignoud
Julien Vignoud
The work pre-refactoring has been left in the code unused, we are removing it in #652. For future use, it can be found in `discojs/discojs-core/src/aggregator/robust.ts` at commit [b9db50f](https://github.com/epfml/disco/blob/b9db50f3e777d8f9cfa343db5ae4c8887314862f/discojs/discojs-core/src/aggregator/robust.ts)
[MNIST web training with ONNX](https://github.com/microsoft/onnxruntime-training-examples/tree/master/on_device_training/web)
@tharvik do you have an opinion on putting the gpt-tfjs test spec file, which relies on tfjs-node, in the `discojs-core/.../gpt`? Or better to put it in discojs-node?
> for now, there isn't a good place to put theses tests, it's de facto in server/tests. with default tasks splitted from discojs-core (tracked in https://github.com/epfml/disco/issues/647), it'll fit naturally in...
@tharvik is this issue still relevant?
Related to #647
I think what I find confusing is that the work "task" is usually used at a higher level in machine learning. For example, the first google result of "machine learning...
Closing as there is no current plan/need to integrate the dataset storage feature and the PR is getting old.
Sounds good! Is there a quick fix to support text dataset? Merging this PR means that wikitext will not be available on discolab.ai until a new PR implements text datasets.
> as no dataset was entered (it's undef), one can't train anything. I added a small message to tell users to enter some. one more functional way would be to...