Closed mjack3 closed 1 year ago
I'm also developing the fastflow model. I have a WIP version that gets near results (- 0.5).
It's difficult replicate because paper has incongruences and It doesn't provide implementation details.
If you wish, we can cooperate
Closing this for now because I don't think we'll implement FastFlow any time soon.
I got a dev branch locally for it, but very difficult to reproduce. Don’t trust the all the paperswithcode results until it is reproducible ;)
On Sun, 10 Apr 2022 at 13:16, Miguel Campos Romero @.***> wrote: