support lm head quantizaiton and export to Intel cpu #76
wenhuach21
merged
57 commits into
main
from
lm-head
fix a bug in example
7d020db2
Merge branch 'main' of https://github.com/intel/auto-round
fbe69d5c
Merge branch 'main' of https://github.com/intel/auto-round
596a18f5
Merge branch 'main' of https://github.com/intel/auto-round
10add8ce
Merge branch 'main' of https://github.com/intel/auto-round
003b60a6
Merge branch 'main' of https://github.com/intel/auto-round
9d495140
Merge branch 'main' of https://github.com/intel/auto-round
3b7f386c
Merge branch 'main' of https://github.com/intel/auto-round
d3f14df6
Merge branch 'main' of https://github.com/intel/auto-round
76e4d909
Merge branch 'main' of https://github.com/intel/auto-round
08e46acb
Merge branch 'main' of https://github.com/intel/auto-round
15b756b0
Merge branch 'main' of https://github.com/intel/auto-round
40425bbd
step 1 for setting up lm-head quantization
9484ffa9
support basic quantization tuning of layers outside blocks
e3703d64
support use_quant_input tuning of layers outside blocks
d6d40d2e
![pre-commit-ci[bot]](https://avatars.githubusercontent.com/in/68672?v=4&s=60)
[pre-commit.ci] auto fixes from pre-commit.com hooks
e38c3fa6
fix some issues
f1a5cf8e
fix some issues
9d8ee5ff
save memory
ded6d615
[pre-commit.ci] auto fixes from pre-commit.com hooks
e1711393
ugly workaround to save lm-head memory
7ad949b4
[pre-commit.ci] auto fixes from pre-commit.com hooks
0cbd77e7
fix one issue
294bc006
fix one issue
051aa435
set to true as a workaround for itrex
0527c66c
set to low_gpu_mem_usage to false as a workaround for itrex
fac2640f
force low_gpu_mem_usage to false as a workaround for itrex
b5e31dc1
add comments
971a36ca
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa987da8
fix bugs
3d3a60e2
[pre-commit.ci] auto fixes from pre-commit.com hooks
2a481eaa
Support XPU export (#77)
9b16ae2a
revert the change
b1456ef8
fix cost time
926ed382
fix the scale dtype, need to revert later
38d90ed3
aligning XPU's export configuration with INC (#79)
366f4ffd
wenhuach21
commented
on 2024-04-19
remove only_quantize_blocks, update xpu shell
87d1ad31
[pre-commit.ci] auto fixes from pre-commit.com hooks
9a570009
fix critical bug in lm head tuning (#82)
aed6ce87
fix some issues
2044d78a
fix some issues
3dbb7b3f
fix one issue
c2da84fb
Merge branch 'main' into lm-head
779ccd2d
add comment
6fc9331b
[pre-commit.ci] auto fixes from pre-commit.com hooks
7a11231d
wenhuach21
changed the title
fix preci issue
624c03a2
refine hpu memory check
45a2e973
follow weight config to decide whether quantizing lm-head or not
32870f82
Merge branch 'lm-head' of https://github.com/intel/auto-round into lm…
6bfb8ab4
[pre-commit.ci] auto fixes from pre-commit.com hooks
cb744d46
fix bug
6d4b9d13
Merge branch 'lm-head' of https://github.com/intel/auto-round into lm…
a2147d76
fix bug and add unit test
36ac0840
[pre-commit.ci] auto fixes from pre-commit.com hooks
6a93a2ea
fix ut
23cd7580
Merge branch 'lm-head' of https://github.com/intel/auto-round into lm…
56cf41c6
[pre-commit.ci] auto fixes from pre-commit.com hooks
9a9a1f17
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub