Detect and fix most `_init_weights()` issues - make it work for composite models #37070
Cyrilvallez
marked this pull request as ready for review 1 year ago
Cyrilvallez
force pushed
from
8574b67a
to
759c5c8b
1 year ago
Cyrilvallez
force pushed
from
1ceba43f
to
f21633f2
1 year ago
Cyrilvallez
force pushed
from
c84f00b5
to
39ddd6e9
1 year ago
Cyrilvallez
changed the title Detect and fix all `_init_weights()` issues Detect and fix most `_init_weights()` issues - make it work for composite models 1 year ago
Cyrilvallez
force pushed
from
13ee1dad
to
8e55a6f1
1 year ago
Cyrilvallez
force pushed
from
6ef2c6da
to
e5d5eccb
1 year ago
Update test_modeling_common.py
47c8e54b
Fix Llama and its modular children
abe04889
Update test_modeling_common.py
4749c0dd
qwen3
49d76252
first try at prioritizing models
80e52d75
Update test_modeling_common.py
155969d8
Update test_modeling_common.py
9a154505
Update test_modeling_common.py
62725b44
test
b5eb6cd0
fix
d0a016a3
fix
dceab888
more models
32518528
more
15bcf974
more
997ba7e2
more
05519112
smarter init for composite models!
eb4bfd3c
fix post rebase
2082647e
smol
9a83073c
fix missing args
4fc6898f
more
fa736c93
typo
7001d130
Super elegant and efficient init for submodels
b0839654
Update modeling_utils.py
7adb98c1
style
bf9b49f7
last fixes
e4141c0a
cleanup
a04f7d59
finalize cleanup
8ed50ab8
CIs
a9c303ec
improve docstring
23eb8c19
Update modeling_utils.py
1aa8914c
llama4
28f86573
style
ce281b83
CIs
c135488c
style
f41f9cc9
add dpt
6f6364ca
granite speech
ce665b86
Cyrilvallez
force pushed
from
bc884aef
to
ce665b86
1 year ago
qwen 2.5 omni
e3ccb5f4
better fix
1034c8c3
Parse the config file instead
25d47e44
CIs
4400c527
Cyrilvallez
deleted the fix-all-init-weights branch 1 year ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub