lucidrains/BS-RoFormer
Implementation of Band Split Roformer, SOTA Attention network for music source separation out of ByteDance AI Labs
PythonMIT
Issues
- 2
- 0
Train Melband on 3090
#38 opened by lucellent - 0
Issues Related to TensorRT Accelerated Inference
#37 opened by bfloat16 - 0
Apple Silicon / MPS support?
#35 opened by dj-nuo - 0
- 2
- 0
Linear Attention
#33 opened by Psarpei - 1
Reproduce the results from the paper
#30 opened by Andrew98998 - 1
Linear Attention temperature initialization
#28 opened by f0k - 1
Flash attention error in Linear Attention layer
#29 opened by ZFTurbo - 16
MelRoformer parameters from paper
#26 opened by ZFTurbo - 5
Gates in Attention module of bs_roformer.py
#27 opened by Psarpei - 5
Problem with DataParallel
#25 opened by ZFTurbo - 4
Hi @lucidrains + @ZFTurbo
#5 opened - 36
Training of BS-RoFormer
#4 opened by ZFTurbo - 7
- 3
Potential bug with `num_stems > 1`
#20 opened by dorpxam - 7
MLP design in MaskEstimator
#15 opened by YoungloLee - 6
- 2
How to use it for audio source separation?
#14 opened by asasas234 - 7
Hop length
#11 opened by Psarpei - 0
- 5
Hidden dim in mask estimation module
#12 opened by Psarpei - 7
MelBand doesn't work with stereo
#10 opened by ZFTurbo - 0
Thank you very much to the teacher for providing such good sound source separation.
#1 opened by ybhka2022 - 2
Flash Attention support
#8 opened by dorpxam - 1
- 1
Default time_transformer_depth
#6 opened by iver56 - 1
Input tensor size
#7 opened by Psarpei - 18
Input data shape
#3 opened by ZFTurbo