huggingface/accelerate

Pull Requests Commits

Fix get_logger kwarg documentation issue (#1222)

bcol23 committed 3 years ago

Verified e7bb060c

Fix bug in loading launch config (#1218)

neumyor committed 3 years ago

Verified a15d3074

FIx TPU gradient state (#1219)

muellerzr committed 3 years ago

Verified 7e7f3445

ds offload optim fix to use CPUAdam (#1208)

pacman100 committed 3 years ago

Verified 10c67463

Fix example in accumulate method (#1211)

VikParuchuri committed 3 years ago

Verified 82c2665c

Fix typo in TPU config (#1202)

muellerzr committed 3 years ago

Verified 2930cac6

Better error message when using multi-GPU and Accelerate on torch <1.9.1 (#1203)

muellerzr committed 3 years ago

Verified 901ab69a

Fix tied weights load (#1204)

sgugger committed 3 years ago

Verified 780e4aa3

Make the Scheduler adjust the steps taken relative to the gradient accumulation steps (#1187)

muellerzr committed 3 years ago

Verified e4620984

Fixup --fsdp (#1198)

muellerzr committed 3 years ago

Verified 017a98c0

[`Accelerator`] We should not call `to` on modules that wraps `accelerate` loaded models (#1172)

younesbelkada committed 3 years ago

Verified d1aa5581

Set drop last to ensure modulo16 restriction for fp8 (#1189)

ksivaman committed 3 years ago

Verified 41479fe4

Only convert linear layers with weights multiple of 16 (#1188)

sgugger committed 3 years ago

Verified eac5d13c

add `use_orig_params` to FullyShardedDataParallelPlugin (#1184)

pacman100 committed 3 years ago

Verified b228136c

Add documentation about PyTorch FSDP state dict behavior (#1181)

VikParuchuri committed 3 years ago

Verified 90deb748

Support special mapping of dtypes when preparing device map (#1179)

sgugger committed 3 years ago

Verified d9427087

fixed typo in launch.py tpu_pod_launcher (#1180)

hackpert committed 3 years ago

Verified 37831808

Add repr to AlignHook for easier debugging. (#1177)

sgugger committed 3 years ago

Verified ea836f30

Run accelerate_test in cli (#1176)

muellerzr committed 3 years ago

Verified a4c94762

Fix CPU error always being raised (#1175)

muellerzr committed 3 years ago

Verified 3ca8c9a9

Fix accelerate test with new config_file errors (#1169)

sgugger committed 3 years ago

Verified 2f83b1af

Fix backward compatibility in configs wrt dynamo backend (#1168)

sgugger committed 3 years ago

Verified b0591c66

sgugger committed 3 years ago

Verified d9871c0f

Remove outdated command directions and use in tests (#1166)

muellerzr committed 3 years ago

Verified abc2beb4

Fix what files get deleted through `total_limit` (#1165)

muellerzr committed 3 years ago

Verified 8749b4ec

Document skip_first_batches in the checkpoint usage guides (#1164)

muellerzr committed 3 years ago

Verified 4a3eaee6

[`Accelerator`] Fix issue with 8bit models (#1155)

younesbelkada committed 3 years ago

Verified 3533e2b0

Attempt to fix import error when PyTorch is build without `torch.distributed` module (#1108)

mfuntowicz committed 3 years ago

Verified 3e0ceac7

Let GradientState know active dataloaders and reset the remainder (#1162)

muellerzr committed 3 years ago

Verified 03b617b6

update support for torch dynamo compile (#1150)

pacman100 committed 3 years ago

Verified 840bb1ae

Newer Older