transformers
`HfQuantizer` class for quantization-related stuff in `modeling_utils.py`
#26610
Merged
Merged
`HfQuantizer` class for quantization-related stuff in `modeling_utils.py` #26610
ArthurZucker
merged
83 commits into
huggingface:main
from
poedator:quant
squashed earlier commits for easier rebase
e0650b27
rm rebase leftovers
42adf9d7
4bit save enabled @quantizers
7f57f26b
TMP gptq test use exllama
f1f5da08
fix AwqConfigTest::test_wrong_backend for A100
a94d3a7b
quantizers AWQ fixes
0b30de46
_load_pretrained_model low_cpu_mem_usage branch
4cdaf0dd
quantizers style
0db11076
remove require_low_cpu_mem_usage attr
89d11771
rm dtype arg from process_model_before_weight_loading
0c71b00e
rm config_origin from Q-config
2b4122a5
rm inspect from q_config
02ad5624
fixed docstrings in QuantizationConfigParser
3e51d51b
logger.warning fix
25693670
mv is_loaded_in_4(8)bit to BnbHFQuantizer
32592434
is_accelerate_available error msg fix in quantizer
ab614178
split is_model_trainable in bnb quantizer class
95e44cda
rm llm_int8_skip_modules as separate var in Q
b936cfb9
Q rm todo
0b40d21d
fwd ref to HFQuantizer in type hint
c53a3fb5
rm note re optimum.gptq.GPTQQuantizer
dbd93f25
quantization_config in __init__ simplified
e34bd581
replaced NonImplemented with create_quantized_param
fcd5a7aa
rm load_in_4/8_bit deprecation warning
954c5e6c
QuantizationConfigParser refactoring
49e163f9
awq-related minor changes
f8b9e07e
awq-related changes
5eaf9acf
awq config.modules_to_not_convert
d678d995
raise error if no q-method in q-config in args
7c9c49b4
minor cleanup
0d739d3a
awq quantizer docstring
b5f2bab1
combine common parts in bnb process_model_before_weight_loading
af334637
revert test_gptq
d4af5f12
.process_model_ cleanup
94f2cc7c
restore dict config warning
ec77d102
removed typevars in quantizers.py
f5b98495
cleanup post-rebase 16 jan
fb37bb89
QuantizationConfigParser classmethod refactor
cdc71c80
rework of handling of unexpected aux elements of bnb weights
e6df6ed3
moved q-related stuff from save_pretrained to quantizers
1c433f57
refactor v1
60781dd1
more changes
842391aa
fix some tests
08034405
remove it from main init
594d1a9f
ooops
a771ab7d
Apply suggestions from code review
aa4ec34e
fix awq issues
53619de2
Merge remote-tracking branch 'upstream/main' into hf-quantizer-work
cd4aa905
fix
a988d016
fix
a911e7d5
fix
38865591
fix
43e5e70b
fix
c1dcaa3e
fix
b0ac4a79
Merge branch 'main' into hf-quantizer-work
1575c474
add docs
ad8d7f63
Apply suggestions from code review
89cf6cf0
Apply suggestions from code review
0ebaf4e7
Update docs/source/en/hf_quantizer.md
adaae054
address comments
f0b5f965
fix
30e1fc28
Merge branch 'hf-quantizer-work' of https://github.com/younesbelkada/…
3b7e6257
fixup
493d117b
Update src/transformers/modeling_utils.py
48c57615
Update src/transformers/modeling_utils.py
3744fb1f
address final comment
c4995ab8
Merge branch 'hf-quantizer-work' of https://github.com/younesbelkada/…
17f95bf2
update
abb4db35
Update src/transformers/quantizers/base.py
7e5a5b88
Update src/transformers/quantizers/auto.py
122b494c
fix
901ace5b
Merge remote-tracking branch 'upstream/main' into hf-quantizer-work
2da5233d
add kwargs update
2ab7fd53
Merge remote-tracking branch 'upstream/main' into HEAD
242682c0
Merge branch 'quant' into hf-quantizer-work
e387f68a
fixup
4c0c33ed
add `optimum_quantizer` attribute
c37b222c
oops
ca40b049
Merge pull request #5 from younesbelkada/hf-quantizer-work
c0ed16a8
rm unneeded file
7a764fb6
Merge branch 'quant' of https://github.com/poedator/transformers into…
377943df
younesbelkada
approved these changes
on 2024-01-30
younesbelkada
changed the title
ArthurZucker
approved these changes
on 2024-01-30
Merge remote-tracking branch 'upstream/main' into HEAD
85d46563
fix doctests
deb7696e
winglian
commented
on 2024-02-02
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub