Adding the new feature of FPDT #6462
loadams
merged
74 commits into
deepspeedai:master
from
YJHMITWEB:master
fix the bug of deepspeed sequence parallel working with batch size la…
c0768279
Merge branch 'master' into master
1b8a8c15
apply yapf formatting
ed34e89f
Formatting fixes
89b119e5
Merge branch 'microsoft:master' into master
7db5798f
add FPDT
0beff247
Merge branch 'master' into master
4522ed7a
delock
commented
on 2024-08-30
delock
commented
on 2024-08-30
delock
commented
on 2024-08-30
inkcherry
commented
on 2024-08-30
tohtana
commented
on 2024-08-30
tohtana
commented
on 2024-08-30
Merge branch 'master' into master
c15d1d85
modify streams
69f38928
modify streams
8ef9f5aa
Merge branch 'master' into master
b43c5ecf
remove duplication of alltoall
a55d1f5d
Merge branch 'master' of github.com:YJHMITWEB/DeepSpeed
1cbd59d5
remove duplication of pos
6bfd76f4
fix format
4eeadca6
Merge branch 'master' into master
8994991a
fix format and add unit test for fpdt
128286ca
Merge branch 'master' of github.com:YJHMITWEB/DeepSpeed
386f606d
add einops
ebea5b03
add flashattn
5c8eec86
Merge branch 'master' into master
a7e175ac
add requirements for flash-attn in FPDT
764a5727
Merge branch 'master' of github.com:YJHMITWEB/DeepSpeed
14e91b88
Merge branch 'master' into master
77dcd383
Merge branch 'master' into master
7a5c29c6
skip test when fa is unavailable
534cb937
formatting
972ddda9
add workflow to run a6000 tests
37bc6942
Merge branch 'master' into FPDT
8f8aaa03
revert world sizes for tests
ac7baf67
Merge pull request #1 from YJHMITWEB/tohtana/merge_FPDT
0d2b6247
update workflow
89355294
loadams
commented
on 2024-11-05
loadams
commented
on 2024-11-05
update image version
edd2e051
remove --no-build-isolation
464d117f
remove requirements file for flash-attn
7389f663
remove flash-attn requirements from setup.py
5f859be9
fix pip command
56cb6476
modify unit test for fpdt
164f459e
modify unit test for fpdt
3eb816db
modify unit test for fpdt
2ae68dc1
modify unit test for fpdt
b1b26880
modify unit test for fpdt
67aa3dfb
loadams
commented
on 2024-11-05
modify unit test for fpdt
42461d27
modify unit test for fpdt
d637d604
modify unit test for fpdt
907c79d0
modify unit test for fpdt
8f5d0397
modify unit test for fpdt
02c2fbf6
modify unit test for fpdt
f570213c
add condition for using fpdt offloading
5b8c419f
add condition for using fpdt offloading
bd090c84
add flash-attn version check
e48e85ba
Merge branch 'master' into master
af24777a
add unit test directory as test trigger
ebaf56c9
loadams
commented
on 2024-11-06
add cron for test and reporting for nightly CI failures
9e811b8d
add multiGPU fpdt unit test
a7522da8
add multiGPU fpdt unit test
209adab4
add multiGPU fpdt unit test
dbeea8a6
add multiGPU fpdt unit test
845e42d5
add multiGPU fpdt unit test
8b2549c9
add multiGPU fpdt unit test
058c973d
add multiGPU fpdt unit test
0dcc234d
add multiGPU fpdt unit test
d1be5d35
add multiGPU fpdt unit test
3a0feba3
add multiGPU fpdt unit test
8c57812c
add multiGPU fpdt unit test
43decf6f
add multiGPU fpdt unit test
d39585c6
add multiGPU fpdt unit test
389b1a3b
add multiGPU fpdt unit test
958f3bf2
add multiGPU fpdt unit test
af025c51
Merge branch 'master' into master
2230377f
Merge branch 'master' into master
76366900
tohtana
approved these changes
on 2024-11-26
Merge branch 'master' into master
10cdc5fa
Merge branch 'master' into master
b2175a45
Merge branch 'master' into master
1e801cee
disabled auto-merge 1 year ago
Manually disabled by user
Manually disabled by user
Login to write a write a comment.
Login via GitHub