Commits

Commits on Oct 10, 2024

Add API to get devices of offload states (#6586)

authored
adec991
reduce setting global variables to reduce torch compile graph breaks (#6541)

authored
d7ca3d8

Commits on Oct 9, 2024

AIO CPU Locked Tensor (#6592)

jomayeri
and
tjruwase
authored
a1f98bd
Clean up prefetched parameters (#6557)

authored
7d751ee
Update version.txt after 0.15.2 release (#6615)

loadams
and
jomayeri
authored
55f7f37
Enabled Qwen2-MoE Tensor Parallelism (TP) inference (#6551)

gyou2021
and
loadams
authored
474a328
Unpin accelerate tests, update lightning with node16 removal. (#6611)
loadams
authored
1062a0c
Rearrange inference OPS and stop using builder.load (#5490)

authored
645639b

Commits on Oct 8, 2024

Handle when `backend` is also in compile_kwargs (#6502)

authored
ca8b1fe
Fix device selection using CUDA_VISIBLE_DEVICES (#6530)

authored
5cbbff4
Improve DS logging control (#6602)
tjruwase
authored
f74ea69
Add llama3.2 vision autotp (#6577)

Yejing-Lai
and
loadams
authored
e97b453
Pin accelerate to fix CI failures/issues (#6610)
loadams
authored
745dd48
Fix SD workflow (#6609)
loadams
authored
00c4b98
Move V100 workflows from cuda 11.1/11.7 to 12.1 (#6607)
loadams
authored
20695b3

Commits on Oct 7, 2024

Add SSF Best practices badge (#6604)
loadams
authored
940887d
Cleanup CODEOWNERS file to be valid (#6603)
loadams
authored
239b83a

Commits on Oct 4, 2024

[ROCm] Fix subprocess error (#6587)
jagadish-amd
authored
b93c7a2

Commits on Sep 27, 2024

Commits on Sep 26, 2024

[XPU] Support DeepNVMe new code structure (#6532)

authored
d45cfd3
fix errors when setting zero3 leaf modules with torch.compile (#6564)

authored
ba58682
Fix gradient accumulation for Z2+offload (#6550)

authored
c85c870
[Accelerator] Cambricon MLU support (#6472)

Andy666G
and
loadams
authored
0fbe96a
DeepNVMe perf tuning (#6560)

tjruwase
and
jomayeri
authored
a540097
Use msgpack for p2p comm (#6547)

tohtana
and
loadams
authored
7622cd9

Commits on Sep 25, 2024

Skip failing newly added tests in accelerate (#6574)
loadams
authored
61de017

Commits on Sep 16, 2024

Added Intel Gaudi to Accelerator Setup Guide (#6543)

authored
2a56f53

Commits on Sep 11, 2024

Add conditional on torch version for scaled_dot_product_attention (#6517)
loadams
authored
170b46e
Avoid security issues of subprocess shell (#6498)

tjruwase
and
loadams
authored
659f6be

Commits on Sep 10, 2024

wrap include cuda_bf16.h with ifdef BF16_AVAILABLE (#6520)
oelayan7
authored
c274839

Commits on Sep 9, 2024

Revert "BF16 optimizer: Clear lp grads after updating hp grads in hook" (#6508)
nelyahu
authored
8fa6b50