torch-native pipeline parallelism for big models #2345
Broken version
e713e28e
Timing I would expect
2767bb19
Working version!
06f04a99
muellerzr
marked this pull request as draft 1 year ago
Use MethodType
9eef9dd4
muellerzr
force pushed
from
66fb6116
to
9eef9dd4
1 year ago
working test
449eb8d9
Tests
77f8e92b
Use no split module classes explicitly
df7779aa
Put split_points in pipelien
e3f6b99b
Store split points in hf_split_points
8792a8c5
fix case num_process=1
7ca4bccf
SunMarc
approved these changes
on 2024-01-24
Allow for dynamic batch padding (#2352)
dac1daa9
Rm literal
364c3b61
Allow users to pass in max_memory
6a8479b2
Note about recursion
303c9cc7
Document, document, document
d497e8af
Right import check
06bbc5b5
Merge branch 'main' into pippy-integration-v2
5e047da8
Fix bug, add tests to multigpu runners
a5059e62
Change default to None
71346a19
kwen2501
approved these changes
on 2024-01-31
SunMarc
approved these changes
on 2024-01-31
Start of docs
fe66b937
Try again?
d2af472b
Try again x2
8dc6c6c8
Trailing comma
4d0aeb2b
Move import
309b71a6
Clean
9f561f11
typehint
d5a6fda3
typo
954a668e
kwen2501
approved these changes
on 2024-02-05
From code review
853f5521
Use num_chunks
1362e5c2
Update tests/test_utils.py
68bd89b3
muellerzr
marked this pull request as ready for review 1 year ago
Bad copy/paste
181fbda9
muellerzr
changed the title Pippy integration v2 torch-native pipeline parallelism for big models 1 year ago
hf_split_points
9157cf16
Apply suggestions from code review
f2c6e088
Year
9f204969
Nit
e1961d6d
better title
8c72a5e6
Rephrase
3eaa9678
Rephrase
31fcde4d
Try spacing maybe?
7c3d1830
muellerzr
merged
0867c093
into main 1 year ago
muellerzr
deleted the pippy-integration-v2 branch 1 year ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub