Tunji Ruwase
|
a1c41e0219
Fix docs
|
1 年之前 |
Tunji Ruwase
|
f638d9225a
Merge branch 'master' of github.com:microsoft/DeepSpeed into olruwase/ds_2921
|
1 年之前 |
Logan Adams
|
869629c210
Add missing RocBlas include (#4557)
|
1 年之前 |
Nadav Elyahu
|
a02de228d0
pipe engine _aggregate_total_loss: more efficient loss concatenation (#4327)
|
1 年之前 |
Michael Wyatt
|
0f2338f7b8
Fix RTD builds (#4558)
|
1 年之前 |
Tunji Ruwase
|
f25ff5b1b6
Docs
|
1 年之前 |
Olatunji Ruwase
|
507fee8471
Merge branch 'master' into olruwase/ds_2921
|
1 年之前 |
Tunji Ruwase
|
d1cefd6155
Detect bf16_optimizer
|
1 年之前 |
Logan Adams
|
e238351101
ROCm 6.0 prep changes (#4537)
|
1 年之前 |
Jeff Rasley
|
488f7e2dd6
[docs] paper updates (#4543)
|
1 年之前 |
Ma, Guokai
|
04cd6af130
turn off I_MPI_PIN for impi launcher (#4531)
|
1 年之前 |
Logan Adams
|
c7724c6181
Switch from HIP_PLATFORM_HCC to HIP_PLATFORM_AMD (#4539)
|
1 年之前 |
Jeff Rasley
|
a5b1cb1eb5
[docs] ZeRO infinity slides and blog (#4542)
|
1 年之前 |
Ramya Ramineni
|
3e4a587135
Added rocblas header (#4538)
|
1 年之前 |
Ilya Vologin
|
beed962c25
[Bug fix] Add rope_theta for llama config (#4480)
|
1 年之前 |
Quentin Anthony
|
8a93ded874
Fixed deepspeed.comm.monitored_barrier call (#4496)
|
1 年之前 |
Liangliang-Ma
|
a7358817f5
fix error type in ccl.py (#4521)
|
1 年之前 |
CurryRice233
|
3e70a88715
Add NPU FusedAdam support (#4343)
|
1 年之前 |
Olatunji Ruwase
|
3b9a384573
Merge branch 'master' into olruwase/ds_2921
|
1 年之前 |
Tunji Ruwase
|
d737cbc7d2
Handle replicated params
|
1 年之前 |
mzl
|
2cbfb89ab2
clear redundant parameters in zero3 bwd hook (#4520)
|
1 年之前 |
Tunji Ruwase
|
51b3af8713
Merge branch 'olruwase/ds_2921' of github.com:microsoft/DeepSpeed into olruwase/ds_2921
|
1 年之前 |
Tunji Ruwase
|
f5c6b2dd6f
PR feedback
|
1 年之前 |
Olatunji Ruwase
|
f21a5de59b
Merge branch 'master' into olruwase/ds_2921
|
1 年之前 |
Tunji Ruwase
|
64d8c0d8ef
Formatting fix
|
1 年之前 |
Tunji Ruwase
|
b13006bc06
Remove logging fix to seperate PR. Relocate conversion script to avoid logging circular import issue
|
1 年之前 |
Olatunji Ruwase
|
3dc989ecc4
Merge branch 'master' into olruwase/ds_2921
|
1 年之前 |
Tunji Ruwase
|
2a60f79301
Enable uni_ckpt for z1
|
1 年之前 |
Jeff Rasley
|
12aedac6ce
add available memory check to accelerators (#4508)
|
1 年之前 |
Sam Ade Jacobs
|
78c518ed97
Update README.md (#4518)
|
1 年之前 |