Add Moonshine #34784
eustlb
merged
93 commits into
huggingface:main
from
eustlb:add-moonshine
config draft
35434da0
full encoder forward
7e180381
full decoder forward
65172519
fix sdpa and FA2
b0efed1e
fix sdpa and FA2
b4d18f9e
moonshine model
b3777e0c
moonshine model forward
e313ab56
fix attention with past_key_values
7a6935a6
add MoonshineForConditionalGeneration
8fda4267
fix cache handling and causality for cross attention
d0ed9172
no causal attention mask for the encoder
461f210e
model addition (imports etc)
22dbaae0
small nit
72ba8c44
nits
f5485043
Update src/transformers/models/moonshine/convert_usefulsensors_to_hf.py
3d52b1e8
add rope_theta
8f82a405
nits
fc73b374
model doc
aedccf5e
xenova
commented
on 2024-12-16
xenova
commented
on 2024-12-16
Update src/transformers/models/auto/configuration_auto.py
09541338
imports
b1f0909d
add MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES
f647a9ff
updates modular
2da255d6
make
407cd3ed
make fix-copies
b3da73da
ruff check examples fix
8cf195d8
fix check_modular_conversion
f78db3d4
Merge branch 'main' into add-moonshine
59278468
nit
b680f0d1
nits
01a4ea17
nits
5124cbd6
copied from -> imports
a528bd38
Merge branch 'main' into add-moonshine
5648c383
imports fix
338c7c0a
integrate attention refacto
2ec366aa
modular edge case
52618866
xenova
commented
on 2024-12-20
remove encoder
467b8e47
convolutions params in config
445dfcbd
run modular_model_converter
a656e8cf
make
65476342
Update docs/source/en/model_doc/moonshine.md
9eec03bb
MoonshineModelTest
53b9b9d1
correct typo
fb2258fa
make style
259c6b32
integration tests
59dd57f2
Merge branch 'main' into add-moonshine
8fe2cdc5
make
02c43438
modular convert
6c7dc79e
sanchit-gandhi
commented
on 2025-01-08
Merge branch 'main' into add-moonshine
89904f9b
stevhliu
approved these changes
on 2025-01-08
name conversion update (up_proj -> fc1 etc)
829a2b86
update config
02a1f1f8
update MLP
80c39cd0
update attention
55ac7e28
update encoder layer
4d540af3
update decoder layer
4f52158b
update convolutions parameters
6568576f
update encoder
495bd735
remove INPUTS_DOCSTRING
48492c25
update decoder
21a743fe
update conditional generation
3b7af33e
update pretrained model
563a3600
imports
0ebd9a59
modular converted
d0233a76
update doc
f9a6cac1
fix
b8a6e3b4
typo
bfea59ae
update doc
ea150d36
update license
8170cc8e
update init
9bc7b353
split config in file
2bbb8eec
two classes for MLP
b42d7f6d
attention from GLM
08272b9a
from GlmRotaryEmbedding
61462dbc
split MLP
c866e585
apply arthur's review suggestions
b0183be3
apply arthur's review suggestions
3aa4f8d4
apply arthur's review suggestions
d64190cd
auto feature extractor
0bfb6bca
convert modular
1268e138
fix + make
141a70ba
Merge branch 'main' into add-moonshine
70f16dcf
convert modular
1f3cc631
make
77ca19af
unsplit config
e6b19db9
use correct checkpoint
7155fbd7
wrap generate
15b2552d
Merge branch 'main' into add-moonshine
89ceca80
update tests
fc4febc5
typos
3e2a84bd
make
9a8f91d9
typo
834c3646
update doc
b491d953
Merge branch 'main' into add-moonshine
d357fffc
Merge branch 'main' into add-moonshine
2a6c59eb
Reviewers
Assignees
No one assigned
Labels
New model
Audio
Milestone
No milestone
Login to write a write a comment.
Login via GitHub