snowfall

snowfall copied to clipboard

Load just manifests

1

@pzelasko I notice with the librispeech example, the path of least resistance when writing scripts is to just call prepare_librispeech() which regenerates the recordings and supervisions' manifests. I think it...

danpovey

Memory blowup

8

Right now we are dealing with an issue in train.py where it uses more and more memory. It seems like stuff isn't getting freed that should be getting freed.

danpovey

Creating minibatch of supervision more efficiently

7

@csukuangfj this is something you could help with. Here, the supervision is created via a loop in Python. Previously we added a function, LinearFsas(), that creates an FsaVec from a...

danpovey