Zhewei Yao
|
0f4f2f982c
Adding DeepSpeed Compression Composer (#2105)
|
2 年之前 |
Jeff Rasley
|
6b9df564df
bump to 0.7.0
|
2 年之前 |
Michael Wyatt
|
ee7ea3b805
use HF NeoX (#2087)
|
2 年之前 |
Aman Sanger
|
9027f861f2
Dont overwrite hook handles in flop profiler (#2106)
|
2 年之前 |
Stas Bekman
|
16699d839f
[ds-inference] checkpoint loading => tqdm (#2107)
|
2 年之前 |
Reza Yazdani
|
aa88137b8d
Add Inference support for running the BigScience-BLOOM Architecture (#2083)
|
2 年之前 |
Jeff Rasley
|
2feaf6d6aa
bump to 0.6.7
|
2 年之前 |
Siddharth Singh
|
c1af73f7c7
Improving memory utilization of Z2+MoE (#2079)
|
2 年之前 |
Zeyu
|
b05237876e
fixed "None type has no len()" (#2091)
|
2 年之前 |
Manuel R. Ciosici
|
db3252b06a
Add missing newline for ZeroOneAdam parameter table (#2088)
|
2 年之前 |
Cheng Li
|
0ad08608a5
remove require grad in params count (#2065)
|
2 年之前 |
Sam Ade Jacobs
|
50a652e757
Codeowner addendum and fix to small model debugging script (#2076)
|
2 年之前 |
Siddharth Singh
|
b3388e1418
Fix partition id in the fp32->fp16 param copying step for z2+cpu-offload (#2059)
|
2 年之前 |
Alex Hedges
|
3540ce74d9
Check for bf16 support only if CUDA is available (#2049)
|
2 年之前 |
Jeff Rasley
|
559fb8e515
[docs] fix broken read-the-docs build (#2075)
|
2 年之前 |
kisseternity
|
9305916d6b
Comments for better understanding of zero stage1_2 (#2027)
|
2 年之前 |
Jeff Rasley
|
9fc4e5f117
add ds inference paper (#2072)
|
2 年之前 |
Quentin Anthony
|
9b70ce56e7
Comms Benchmarks (#2040)
|
2 年之前 |
Alex Hedges
|
76ea0534c1
Fix missing import in replace_module.py (#2050)
|
2 年之前 |
Siddharth Singh
|
38a00bee9e
correct partition_id in fp32 param -> fp16 param for MoE+z2 (#2058)
|
2 年之前 |
Michael Wyatt
|
7bae53d154
Fix for AMD unit tests (#2047)
|
2 年之前 |
Reza Yazdani
|
a04480e192
Fix the half-precision version of CPU-Adam (#2032)
|
2 年之前 |
Conglong Li
|
ff87c4e1f1
Add compression papers (#2042)
|
2 年之前 |
Michael Wyatt
|
5218177922
fixed print statement (#2038)
|
2 年之前 |
Olatunji Ruwase
|
678c3fe330
Split parameter offload from z3 (#2009)
|
2 年之前 |
Olatunji Ruwase
|
2a1a409644
Retain available params until last use (#2016)
|
2 年之前 |
Michael Wyatt
|
ec1ec204c8
Fix inference unit test import error catching (#2024)
|
2 年之前 |
Karim Foda
|
735406e536
fix import errors (#2026)
|
2 年之前 |
Olatunji Ruwase
|
d86a2de993
Use partition size (#2011)
|
2 年之前 |
Quentin Anthony
|
c87f6ee209
DeepSpeed Monitor Module (Master) (#2013)
|
2 年之前 |