transformers
[ `gemma`] Adds support for Gemma 💎
#29167
Merged

[ `gemma`] Adds support for Gemma 💎 #29167

ArthurZucker merged 142 commits into main from add-golden-gate
ArthurZucker
ArthurZucker inital commit
7434ea28
ArthurZucker update
61659253
ArthurZucker update conversion checkpoint
69006fa5
ArthurZucker update conversion script
cb801996
ArthurZucker nits
08252ce2
younesbelkada some fixes
32ea5fb1
ArthurZucker nits
888ab957
ArthurZucker Merge branch 'add-golden-gate' of https://github.com/huggingface/new-…
ce0aa57e
ArthurZucker merge
e303e482
younesbelkada fix permute
aefc4bcd
ArthurZucker nits
78de9f5f
younesbelkada Merge branch 'add-golden-gate' of https://github.com/huggingface/new-…
fb2917df
younesbelkada fix
f3ad1b8f
ArthurZucker nits
e3a0bbd1
ArthurZucker nits
bbee069c
ArthurZucker nits
e0e06463
ArthurZucker fix rope
5f63b4c2
ArthurZucker fix both rope
4734805b
ArthurZucker merge
9b27edd1
ArthurZucker nites
d20574d7
ArthurZucker style
3f76b2ff
ArthurZucker make sure flax works
3449b4bb
ArthurZucker fix flax init code
fbbe1497
ArthurZucker fix foward
bf8ed52f
ArthurZucker nits
8273aa7f
ArthurZucker print flax generation out
cf6345c5
ArthurZucker current code
6f49e21f
ArthurZucker nits
e2b09eed
ArthurZucker SIIIIIIIIIIIIIIIIIII
94e1020a
ArthurZucker update
afd89e85
ArthurZucker add new tokenizer
f21564be
ArthurZucker correct fast tokenizer
5c85aa32
ArthurZucker fix conversion
342095bb
ArthurZucker more comments
b6b6f356
ArthurZucker fix modeling and conversion
f16f9b00
ArthurZucker nits and nits
62077c32
ArthurZucker nits testing
274aea77
ArthurZucker add some tokenization tests
9e629dc9
ArthurZucker add some edge cases
b128ceba
younesbelkada add slow tests and fix them
7bec5c60
younesbelkada fixup
51872ef8
younesbelkada fix copies for modeling
c8f89d24
younesbelkada fix copies
367cdcdc
younesbelkada add 7B slow tests
9fc4fd8a
younesbelkada fix
ea24f257
younesbelkada fix
e337cf2e
younesbelkada fix tests
b3565618
ArthurZucker make tokenizer cis go green
423d0b1e
ArthurZucker styling
ac8ee3d3
ArthurZucker last tokenizer nits
ee4eebfd
ArthurZucker update jax tests
06d3ff7f
ArthurZucker fix flax for 7b
b7b31cac
ArthurZucker add jit testing 🤗
7626e71c
ArthurZucker cleanups
5db6ddc5
ArthurZucker isolated nit, inv_freq for rotary_emb.inv_freq
5123d3cd
ArthurZucker propagate to jax
ca759151
ArthurZucker Apply suggestions from code review
36ecdc96
younesbelkada adjust test
792dcf01
younesbelkada Merge branch 'add-golden-gate' of https://github.com/huggingface/new-…
d700211e
younesbelkada fix conversion script
dbc976ef
ArthurZucker change name
10ace9f0
ArthurZucker correct file names
6ff459c6
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/new-model-ad…
a5bb7a21
ArthurZucker update conversion script
2126d1ca
pcuenca Fix bos and eos token ids in the model configuration (#3)
ac7ac879
ArthurZucker update modelling
bd2a7600
ArthurZucker update conversion script
123d72c2
ArthurZucker merge
aab31105
younesbelkada Merge remote-tracking branch 'gg/main' into HEAD
0d2acffc
younesbelkada add static cache for gemma
4e08dd9c
younesbelkada fix sdpa generate
6976cab7
younesbelkada fix batched
92c7a7f1
younesbelkada multiple fixes
f453a42e
younesbelkada fix FA2
491ea9ff
younesbelkada final fix
4b4b621f
pcuenca Rename a few missing strings and filenames (#4)
e9d36beb
younesbelkada Merge branch 'main' into add-golden-gate
cdf93334
younesbelkada merge with upstream main
51a7d9f3
younesbelkada fix copies
89384e83
younesbelkada fix copies
648ef042
younesbelkada fix fixup
fd35d326
younesbelkada fix fixup
58d71465
younesbelkada fix
14d0f545
younesbelkada fix
3822c4c5
younesbelkada final tests
e50721d9
fix fx gemma tests
a7645433
fix fx bf16/fp16 tests
32a04921
update slow fx tests
069b8b57
fx slow tests: one logits, one generation
154e1df6
move jit test standalone
a4074b03
ArthurZucker Merge commit '0996a10077219de0556281511fc02f3ab68002d5' into add-gold…
d7c2eb2d
ArthurZucker Apply suggestions from code review
fc6ac3b7
ArthurZucker nits
a259c041
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/new-model-ad…
c590a9a8
ArthurZucker tokenizer updates
b8d8edbf
ArthurZucker more tokenization updates: custom GemmaSentencepieceExtrator
dc9ecf69
ArthurZucker style
1a381bd8
ArthurZucker Update src/transformers/cache_utils.py
f7496a48
ArthurZucker Update src/transformers/models/gemma/__init__.py
170832a3
ArthurZucker Update tests/models/gemma/test_modeling_flax_gemma.py
b32adc0d
ArthurZucker small nits
24d5191f
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/new-model-ad…
6d5a2feb
ArthurZucker style
403c796f
ArthurZucker update tokenization test
3633bb07
ArthurZucker fix the rotary embedding
df865806
ArthurZucker with style
565028aa
younesbelkada fix slow tests
362d2cda
ArthurZucker WARNING this commit might be very important for precisions
fb87315a
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/new-model-ad…
b63d2e00
ArthurZucker Update tests/models/gemma/test_modeling_flax_gemma.py
9b834f76
ArthurZucker Update src/transformers/models/gemma/configuration_gemma.py
e33fb3d0
ArthurZucker Update src/transformers/models/gemma/modeling_flax_gemma.py
0aa94f11
ArthurZucker small nits here and there!
e06809c2
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/new-model-ad…
eb2c69dc
ArthurZucker forgotten nit
db63274e
ArthurZucker remove on the fly computation of inv_freq
275ee4a2
ArthurZucker revert previous change, let's be safe and for now re-compute freq cis…
6d81a998
younesbelkada Apply suggestions from code review
c9a4d864
younesbelkada Update src/transformers/models/gemma/convert_gemma_weights_to_hf.py
1c42d4a3
younesbelkada Update src/transformers/models/gemma/convert_gemma_weights_to_hf.py
9c4d4391
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
a56f5554
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
03c8e8f5
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
f37f517c
younesbelkada Update tests/models/gemma/test_modeling_flax_gemma.py
e5bfdd0e
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
d83e0988
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
09717b6b
younesbelkada Update tests/models/gemma/test_tokenization_gemma.py
cce69c0a
younesbelkada Update tests/models/gemma/test_tokenization_gemma.py
dde30a5f
younesbelkada Update tests/models/gemma/test_tokenization_gemma.py
02a2d38b
younesbelkada Update tests/models/gemma/test_tokenization_gemma.py
c975dae8
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
5cf4da3c
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
f1980155
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
ac82e00a
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
7253c9ff
younesbelkada Update tests/models/gemma/test_modeling_gemma.py
bc7ebaf7
ArthurZucker nit conversion script link
4ad0d523
younesbelkada
younesbelkada approved these changes on 2024-02-21
younesbelkada fix some tests
7db13fad
ArthurZucker add not doctest and pr doctest
60f8ba6a
ArthurZucker Merge branch 'add-golden-gate' of github.com:huggingface/transformers…
1bf51b30
ArthurZucker repo consistency
2a4d326f
ArthurZucker fix last CIs 🚀
ea9eb109
ArthurZucker ArthurZucker marked this pull request as ready for review 2 years ago
ArthurZucker ArthurZucker added New model
ArthurZucker update all readmes
556f743b
HuggingFaceDocBuilderDev
ArthurZucker ArthurZucker merged 594c1277 into main 2 years ago
ArthurZucker ArthurZucker deleted the add-golden-gate branch 2 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone