huggingface/accelerate

Pull Requests Commits

Adds tag v0.27.2 for pypi

muellerzr committed 1 year ago

5eb7ab9f

Fix seedable sampler logic and expound docs (#2434)

muellerzr committed 1 year ago

5114a37e

Release: v0.27.1

muellerzr committed 1 year ago

ad8a2db6

Release: v0.27.0

muellerzr committed 1 year ago

b7087be5

[Fix] make all tests pass on XPU (#2427)

faaany committed 1 year ago

Verified f75c6245

bug fix (#2426)

faaany committed 1 year ago

Verified 9c1d5bac

Fix fp8 things (#2403)

muellerzr committed 1 year ago

Verified b0b867da

[FIX] fix the wrong `nproc_per_node` in the multi gpu test (#2422)

faaany committed 1 year ago

Verified 433d693b

Migrate pippy examples over and run tests (#2424)

muellerzr committed 1 year ago

Verified c3aec59b

Make output end up on all GPUs at the end (#2423)

muellerzr committed 1 year ago

Verified 9467a627

Update FSDP docs (#2430)

pacman100 committed 1 year ago

Verified 86228e32

muellerzr committed 1 year ago

06b138d8

torch-native pipeline parallelism for big models (#2345)

muellerzr committed 1 year ago

Verified 0867c093

Use Ruff for formatting too (#2400)

akx committed 1 year ago

Verified 0e1ee4b9

muellerzr committed 1 year ago

Verified d8a64cb7

Adding Local SGD support for NPU (#2415)

ji-huazhong committed 1 year ago

Verified b703efdc

Fix the size of int and bool type when computing module size (#2411)

notsyncing committed 1 year ago

Verified 68f54720

Fix XPU inference (#2383)

notsyncing committed 1 year ago

Verified 46f1391b

Added activateEnviroment.sh to readme (#2409)

TJ-Solergibert committed 1 year ago

Verified cd7ff5e1

Fix CI due to pytest (#2408)

muellerzr committed 1 year ago

Verified f4b411f8

Revert "[don't merge yet] unpin torch (#2406)" (#2407)

muellerzr committed 1 year ago

Verified 7ba64e63

[don't merge yet] unpin torch (#2406)

ydshieh committed 1 year ago

Verified 8b770a7d

Address PIP-632 deprecation of distutils (#2388)

AieatAssam committed 1 year ago

Verified 3d8b998f

Pin torch version (#2394)

Rocketknight1 committed 1 year ago

Verified 03365a3d

Fix batch_size sanity check logic for `split_batches ` (#2344)

izhx committed 2 years ago

Verified 7aafa256

device agnostic cli/data_loader/grad_sync/kwargs_handlers/memory_utils testing (#2356)

wangshuai09 committed 2 years ago

Verified f88661b5

Add `adapter_only` option to `save_fsdp_model` and `load_fsdp_model` to only save/load PEFT weights (#2321)

AjayP13 committed 2 years ago

Verified 581fabba

modified big_modeling.py (#2376)

BlackSamorez committed 2 years ago

Verified e909eb34

`add_hook_to_module` and `remove_hook_from_module` compatibility with fx.GraphModule (#2369)

fxmarty committed 2 years ago

Verified 7644a02e

device agnosic optimizer testing (#2363)

ji-huazhong committed 2 years ago

Verified 162a8216

Older