add zero12 support
87f41420
add support for fs w. fp16/z123, plus z3 linear fix
876064cb
cleanup
9c7a82dc
jeffra
marked this pull request as draft 3 years ago
jeffra
changed the title Fairseq support [WIP] Fairseq support 3 years ago
Merge branch 'master' into jeffra/fs-v062
0003660d
remove pytest tmp
537df2ed
add base class to all deepspeed optimizer wrappers
b2bbf60c
mrwyattii
approved these changes
on 2022-04-29
Merge branch 'master' into jeffra/fs-v062
ad30db14
fix issue w. attr passthrough, nn.module attrs were not passed
4a97bc8e
add model_f to pipe load_module_state_dict
e5b2df81
register client model in nn.module + other clean-up
143a348e
add lr override support to fused fp16 optim
271e509d
use the loss fn directly from the module
5ecad475
set client model when amp is enabled
beb3cea2
Merge branch 'master' into jeffra/fs-v062
4f7f1883
jeffra
marked this pull request as ready for review 3 years ago
jeffra
changed the title [WIP] Fairseq support Fairseq support 3 years ago
cli99
approved these changes
on 2022-05-06
Merge branch 'master' into jeffra/fs-v062
36ec58f0
skip zero tiled linear tests for now
f95a1e4d
Merge branch 'master' into jeffra/fs-v062
62b00411
Merge branch 'master' into jeffra/fs-v062
52745f0d
tjruwase
requested changes
on 2022-05-10
rename model_f -> custom_load_fn
e9a5289b
jeffra
merged
50893458
into master 3 years ago
mrwyattii
deleted the jeffra/fs-v062 branch 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub