Releases: lucidrains/musiclm-pytorch
Releases · lucidrains/musiclm-pytorch
0.1.1
precompute audio and text layers to be used for hierarchical contrast…
0.1.0
add the hierarchical contrastive loss from the ViCHA strategy in http…
0.0.28
allow for sampling multiple pieces of music and selecting top match w…
0.0.26
map location to cpu when loading state dict
0.0.25
fix save command not using the one on the trainer class, for saving o…
0.0.24
0.0.24
0.0.23
add qk rmsnorm
0.0.22
avoid a division
0.0.21
repeat batch for relative positional bias only if doing patch dropout
0.0.20
add best relative positional encoding for AST