tsai

tsai copied to clipboard

I am wondering if you are interested in implementing TS2Vec(https://arxiv.org/abs/2106.10466). This is pretty new and they claimed that they had achieved better results on USC datasets than TST , Inception and many other current state of art models. The author implemented the code in https://github.com/yuezhihan/ts2vec. BTW, Happy Holidays!

It sounds very interesting indeed. It's one of the models I'd certainly like to adapt to tsai (since it's already implemented). I'll read the paper in detail and review the code to see how much effort it'd take to add it to tsai.

great. I looked at their code before, the model part should be straightforward, but they defined two their own loss functions, not sure where are the best places to adjust in tsai.

Michael

在 2021年12月24日，19:11，Ignacio Oguiza @.***> 写道：

 It sounds very interesting indeed. It's one of the models I'd certainly like to adapt to tsai (since it's already implemented). I'll read the paper in detail and review the code to see how much effort it'd take to add it to tsai.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.

Closed due to lack of activity.