Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
This project is mirrored from
https://github.com/lucidrains/audiolm-pytorch
. Pull mirroring updated
Sep 19, 2024
.
0.2.3
bcf77f91
·
fix issue with accelerator wrapped models when saving and loading
·
Jan 04, 2023
0.2.2
e4d31ce4
·
actually save optim state for soundstream
·
Jan 04, 2023
0.2.1
89a35002
·
add helper save load methods to all the trainers, for saving optimizer states,...
·
Jan 04, 2023
0.2.0
deceb7a0
·
bring the feedforward design from localvit to audiolm
·
Dec 27, 2022
0.1.22
94fe05d1
·
0.1.22
·
Dec 15, 2022
0.1.21
d4e278b6
·
0.1.21
·
Dec 15, 2022
0.1.20
b0384e83
·
fix soundstream training
·
Dec 14, 2022
0.1.19
8ebbb2e9
·
fix bug with collation
·
Dec 14, 2022
0.1.18
60e4cd84
·
not all the transformer may need conditioning, perhaps only semantic is...
·
Dec 13, 2022
0.1.17
ca13fbae
·
get a working example of text conditioned semantic transformer training
·
Dec 13, 2022
0.1.16
de27071b
·
switch to modrelu, seems to be recommended by a lot of torch complex packages
·
Dec 12, 2022
0.1.15
aa053305
·
allow fine transformer to be trained on pre-encoded soundstream codebook indices
·
Dec 11, 2022
0.1.14
8ad550a5
·
set things up for eventual text conditioned audio synthesis
·
Dec 09, 2022
0.1.12
a1cf75ca
·
expose the ema decay for the residual vq in soundstream
·
Dec 08, 2022
0.1.11
03b21607
·
0.1.11
·
Dec 08, 2022
0.1.10
1a6cfc30
·
0.1.10
·
Dec 07, 2022
0.1.9
d0151be9
·
some tweaks to soundstream, include commitment loss
·
Dec 07, 2022
0.1.7
98723add
·
validate transformers being passed into audiolm
·
Dec 07, 2022
0.1.6
25c5ce6c
·
only log errors
·
Dec 07, 2022
0.1.5
dd3dcdbf
·
quick validation for vq-wav2vec
·
Dec 06, 2022
Prev
1
…
9
10
11
12
13
14
15
16
17
Next