search

Maghoumi / pytorch-softdtw-cuda

Fast CUDA implementation of (differentiable) soft dynamic time warping for PyTorch
MIT License
615 stars 58 forks source link

package in pypi or conda #20

Open toinsson opened 2 years ago

toinsson commented 2 years ago

Hello,

thanks for the repository, it is very useful to have a soft-dtw algorithm available for GPU computing.

I am wondering whether you are interested in pushing this to pypi and/or conda for ease of use. This might requires some restructuring of the code as a package.

toinsson commented 2 years ago

Hello @Maghoumi,

I have started another project (https://github.com/toinsson/pysdtw/) based on your repo. This allowed me to modify a bit the structure and quickly add some improvements/features.

There is support for packed sequences and long sequences (> 1024) in GPU. These two might be interesting to merge but will change the code significantly. Feel free to indicate whether you want some of these as pull requests.

I also wrote some unittests to make sure both repo are still producing the same results.

Maghoumi commented 2 years ago

Hi @toinsson

Sorry for delays in my responses and PR reviews, I've been tied up with my day job over the past few weeks. It's great to see you took initiative to add those improvements!

I will try out your project as soon as I get some free time (next couple of weeks or so). Very curious to see the performance on packed sequences.