transformers
add VITS model
#24085
Merged
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Overview
Commits
145
Changes
View On
GitHub
add VITS model
#24085
sanchit-gandhi
merged 145 commits into
huggingface:main
from
hollance:vits
hollance
added
New model
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
requested a review
from
Vaibhavs10
2 years ago
hollance
requested a review
from
sanchit-gandhi
2 years ago
Vaibhavs10
commented on 2023-06-23
sanchit-gandhi
approved these changes on 2023-06-25
ArthurZucker
commented on 2023-06-26
hollance
force pushed
2 years ago
hollance
force pushed
2 years ago
hollance
changed the title
[WIP] add VITS model
add VITS model
2 years ago
hollance
marked this pull request as ready for review
2 years ago
hollance
requested a review
from
sgugger
2 years ago
sgugger
commented on 2023-06-27
hollance
force pushed
2 years ago
sanchit-gandhi
commented on 2023-06-28
hollance
force pushed
2 years ago
sanchit-gandhi
approved these changes on 2023-06-29
hollance
force pushed
2 years ago
sanchit-gandhi
commented on 2023-06-29
hollance
requested a review
from
sgugger
2 years ago
sanchit-gandhi
requested a review
from
amyeroberts
2 years ago
amyeroberts
commented on 2023-07-04
ArthurZucker
commented on 2023-07-07
sanchit-gandhi
force pushed
2 years ago
sanchit-gandhi
requested a review
from
amyeroberts
2 years ago
amyeroberts
commented on 2023-08-21
add VITS model
2d539508
let's vits
a0160b1a
finish TextEncoder (mostly)
03a9a6e6
rename VITS to Vits
3b4a42e9
add StochasticDurationPredictor
f3d5db36
ads flow model
ac4f51ae
add generator
e235ec75
correctly set vocab size
da79eb06
add tokenizer
4c3429c7
remove processor & feature extractor
19a4d1bb
add PosteriorEncoder
4e2d98c1
add missing weights to SDP
fb1d546c
also convert LJSpeech and VCTK checkpoints
eef58ee3
add training stuff in forward
d0669c89
add placeholder tests for tokenizer
8aa791bf
add placeholder tests for model
ea980d09
starting cleanup
a648dfb0
let the great renaming begin!
c2a5478f
use config
bba3c888
global_conditioning
3dd078ea
more cleaning
71bcb438
renaming variables
72b2df24
more renaming
ba7cd6f5
more renaming
5d0577a1
it never ends
e8ebd237
reticulating the splines
2ad7b5eb
more renaming
6bbd8a7a
HiFi-GAN
7fc673cc
doc strings for main model
a04a9059
fixup
0afb73e6
fix-copies
fc3d7653
don't make it a PreTrainedModel
b979458d
fixup
1140cc45
rename config options
f5825f88
remove training logic from forward pass
1054dc75
simplify relative position
0cd0ff27
use actual checkpoint
c67696e3
style
fe71af56
PR review fixes
770800d6
more review changes
4a35af08
fixup
9c8d84cb
more unit tests
fd2bba06
fixup
e6e747a8
fix doc test
11b20bcd
add integration test
21c50522
improve tokenizer tests
455a46b0
add tokenizer integration test
65bba351
fix tests on GPU (gave OOM)
31764398
conversion script can handle repos from hub
41e8f336
add conversion script for all MMS-TTS checkpoints
f10a1c4d
automatically create a README for the converted checkpoint
25b4cb96
small changes to config
e784121d
push README to hub
c7220bf3
only show uroman note for checkpoints that need it
2f641d49
remove conversion script because code formatting breaks the readme
c3fec016
make WaveNet layers configurable
66ad4dfe
rename variables
be87976d
add conversion script for all MMS-TTS checkpoints
f10a1c4d
small changes to config
e784121d
push README to hub
c7220bf3
remove conversion script because code formatting breaks the readme
c3fec016
simplifying the math
da62c1fa
output attentions and hidden states
e4691552
also got rid of the other flip
f0e3d8f1
raise error when phonemizer missing
dbe6cecb
update fused tanh sigmoid
f64bd018
reduce dims in tester
c70d8c8b
fix return type
d974893f
all nn's to accept a config
fb642618
make style
2fef806a
remove 'fake' padding token
35ea7d15
harden tokenizer tests
d5b1f5a1
ron norm test
38e901b3
fprop / save tests deterministic
a4d8cf6b
move uroman to tokenizer as much as possible
5089817b
better logger message
4885e0bf
fix vivit imports
c8ead9b6
add uroman integration test
24b27434
make style
26333554
up
a6c8060e
matthijs -> sanchit-gandhi
36ad9eb8
fix tokenizer test
dc6767d5
make fix-copies
b65a3ece
fix dict comprehension
964ca325
fix config tests
1b816adf
fix model tests
56082206
make outputs consistent with reverse/not reverse
7465fb87
fix key concat
2e13470e
more model details
bfa35741
add author
fda2632f
return dict
cfa52ce5
speaker error
38f2caad
labels error
9cbd6893
Apply suggestions from code review
7c6805c9
Update src/transformers/models/vits/convert_original_checkpoint.py
a2513d1d
remove uromanize
e77c6b06
add docstrings
c1561ff1
add docstrings for tokenizer
09669434
upper-case skip messages
46df4b10
fix return dict
1a92ca7d
style
e4a73030
finish tests
1a1edbcc
update checkpoints
36d37588
make style
9289fe4a
remove doctest file
3f972860
revert
e6e80d0d
sanchit-gandhi
force pushed
to
e6e80d0d
2 years ago
fix docstring
6c2ec8eb
fix tokenizer
03ba7869
amyeroberts
approved these changes on 2023-08-25
remove uroman integration test
d9900564
add sampling rate
2e9238ec
fix docs / docstrings
72aa49cf
style
054df38d
Merge branch 'main' into vits
ce52b98c
add sr to model output
acacd4b4
Merge remote-tracking branch 'origin/vits' into vits
8f3e5ebe
fix outputs
6b63f1a1
style / copies
df374f8f
fix docstring
6a367845
fix copies
d2414e7c
remove sr from model outputs
54261a6a
Update utils/documentation_tests.txt
ff3b08c3
add sr as allowed attr
8b01633b
Merge remote-tracking branch 'origin/vits' into vits
5004f42d
sanchit-gandhi
merged
4ece3b94
into main
2 years ago
Login to write a write a comment.
Login via GitHub
Reviewers
sanchit-gandhi
ArthurZucker
amyeroberts
sgugger
Vaibhavs10
Assignees
No one assigned
Labels
New model
Milestone
No milestone
Login to write a write a comment.
Login via GitHub