[Quantization] Quanto quantizer #29023
SunMarc
merged
58 commits into
huggingface:main
from
SunMarc:quanto_integration
start integration
ba4c2b9d
fix
dc88d4f0
younesbelkada
commented
on 2024-02-15
add and debug tests
ee1ee858
update tests
4c50c4d3
make pytorch serialization works
97951ab8
compatible with device_map and offload
c8436cab
fix tests
9d15653a
Merge remote-tracking branch 'upstream/main' into quanto_integration
194a58a9
make style
d1ccb234
add ref
b0e8adbc
Merge remote-tracking branch 'upstream/main' into quanto_integration
61932898
guard against safetensors
b5501576
add float8 and style
29eee507
fix is_serializable
26fe4407
Fix shard_checkpoint compatibility with quanto
6d4ab4c6
more tests
daaeb914
docs
565e699d
adjust memory
56ba7066
better
9329a072
style
9da4d0bd
pass tests
c13a4efc
Update src/transformers/modeling_utils.py
de9c79a4
add is_safe_serialization instead
1a7721ad
Merge branch 'quanto_integration' of https://github.com/SunMarc/trans…
849448d0
Update src/transformers/quantizers/quantizer_quanto.py
c980409d
add QbitsTensor tests
80a5c299
fix tests
c6f66f03
simplify activation list
7deb6448
Update docs/source/en/quantization.md
693e5939
better comment
528916b8
Update tests/quantization/quanto_integration/test_quanto.py
7a95507f
Update tests/quantization/quanto_integration/test_quanto.py
d60c797d
dacorvo
approved these changes
on 2024-03-05
Merge branch 'quanto_integration' of https://github.com/SunMarc/trans…
b73c5ee5
Merge branch 'main' into quanto_integration
1489a1b8
find and fix edge case
5e98443e
Update docs/source/en/quantization.md
850f5e4e
pass weights_only_kwarg instead
5fc659ce
fix shard_checkpoint loading
15f7a2a9
simplify update_missing_keys
bf5f7e6d
Merge remote-tracking branch 'upstream/main' into quanto_integration
c52b6c11
ArthurZucker
approved these changes
on 2024-03-08
Update tests/quantization/quanto_integration/test_quanto.py
ad012e03
recursion to get all tensors
3419a3c2
Merge branch 'quanto_integration' of https://github.com/SunMarc/trans…
bb7c2264
block serialization
a1b3c18d
skip serialization tests
0030d0a2
fix
6d1bce3b
change by cuda:0 for now
e677a536
fix regression
e005baf7
younesbelkada
approved these changes
on 2024-03-14
Merge remote-tracking branch 'upstream/main' into quanto_integration
dc8547de
update device_map
229e4391
fix doc
8f5c9f72
add noteboon
d4cc911f
update torch_dtype
95f05a44
Merge branch 'quanto_integration' of https://github.com/SunMarc/trans…
058937c6
update doc
5bfa654d
typo
b0b79f08
typo
e389cd95
remove comm
46aae3f5
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub