microsoft/DeepSpeed

Pull Requests Commits

[create-pull-request] automated change

loadams committed 285 days ago

cd81fcad

Allow accelerator to instantiate the device (#5255)

nelyahu committed 285 days ago

Verified eb07d41f

Update xpu-max1100.yml with new config and add some tests (#5668)

Liangliang-Ma committed 285 days ago

Verified 4ba49dda

Z3: optimizations for grad norm calculation and gradient clipping (#5504)

nelyahu committed 285 days ago

Verified 6eed634e

Add accelerator setup guides (#5827)

delock committed 285 days ago

Verified 19b01e1d

HPUAccelerator: remove support in set_visible_devices_envs (#5929)

nelyahu committed 286 days ago

Verified a8d1b44d

Update LR scheduler configuration (#5846)

xiyang-aads-lilly committed 286 days ago

Verified f994fb2c

Fix missing ds_id bug (#5824)

tjruwase committed 286 days ago

Verified e3177de6

Update container version for Gaudi2 CI (#5937)

raza-sikander committed 286 days ago

Verified 051c9933

[Zero2] Reduce the unnecessary all-reduce when tensor size is 0. (#5868)

ys950902 committed 286 days ago

Verified 24d1d86b

Use `torch.nan_to_num` replace numpy wrapper one (#5877)

jinyouzhi committed 286 days ago

Verified 862aff37

Log operator warnings only in verbose mode (#5917)

tjruwase committed 286 days ago

Verified 0f2d485c

OptimizedLinear updates (#5791)

jeffra committed 286 days ago

Verified 6e5d58d2

Upgrade HPU image to v1.16.2. (#5610)

vshekhawat-hlab committed 286 days ago

Verified 1890b814

Fix the bug of deepspeed sequence parallel working with batch size larger than 1 (#5823)

YJHMITWEB committed 292 days ago

Verified ffe0af23

Add Japanese translation of DeepNVMe blog (#5845)

tohtana committed 294 days ago

Verified ade7149d

Update clang-format version from 16 to 18. (#5839)

loadams committed 294 days ago

Verified 297a6840

Fix docs building guide (#5825)

tjruwase committed 295 days ago

Verified 0584689d

Fix NV references (#5821)

tjruwase committed 298 days ago

Verified 2ef82232

Non-reentrant checkpointing hook fix (#5781)

ic-synth committed 298 days ago

Verified 3a2d526e

Link GDS blog to site (#5820)

tjruwase committed 299 days ago

Verified 029bb527

Use accelerator to replace cuda in setup and runner (#5769)

Andy666G committed 299 days ago

Verified 3c490f9c

Add fp16 support of Qwen1.5MoE models (A2.7B) to DeepSpeed-FastGen (#5403)

ZonePG committed 299 days ago

Verified 249c1db2

Update to ROCm6 (#5491)

loadams committed 299 days ago

Verified 23d0e022

sequence parallel with communication overlap (#5691)

inkcherry committed 299 days ago

Verified 17ed7c77

[ROCm] Get rocm version from /opt/rocm/.info/version (#5815)

rraminen committed 299 days ago

Verified f82d0886

GDS AIO Blog (#5817)

jomayeri committed 299 days ago

Verified 324ee65c

Fix CPU Adam JIT compilation (#5780)

lekurile committed 299 days ago

Verified 681be6f5

Find ROCm on Fedora (#5705)

trixirt committed 300 days ago

Verified 550f9c75

Fix op_builder for CUDA 12.5 (#5806)

keshavkowshik committed 300 days ago

Verified 08598dbb

Older