GitLab

[FSDP] ssd_offload fixing backward path (grad_fn) for SsdFlatParameter and SsdFlatParameterView (#974)

* [FSDP] fixing backward path for SsdFlatParameter and SsdFlatParameterView when overriding .data

* Get ssd_offload unit tests passing

* [FSDP] get all test_fsdp_offload tests passing w/ ssd_offload on

* Update changelog

Co-authored-by: Min Xu <min.xu.public@gmail.com>

* [chore] update nightly version

* use yesterday's
Co-authored-by: Min Xu <min.xu.public@gmail.com>

Improvements to ssd_offload to support pickling/unpickling SsdTensorHandle (and derived classes) (#964)

Verified that FSDP wrapped models using ssd_offload checkpoint save and restore correctly

This is no longer needed since isort's version is 5.10

Also fix black version to 22.3.0 to fix issue with click
dependency.

Update files that now fail with new version of black {a = 2 ** 4} ->
{a = 2**4}

* made gradient predivide factor configurable

* fix lints
Co-authored-by: Your Name <you@example.com>

* [chore] 0.4.6 release

* added the third party libs removed by precommit

* copyright headers

* isort and pyproject.toml

* precommit and requirement for isort-seed-config

* mypy

* dummy change

* numpy version for pre-commit

* fix mypy issue caused by numpy
Co-authored-by: Min Xu <min.xu.public@gmail.com>

* Adding ELI5 video to Fairscale

* docs: add GH button in support of Ukraine

## Summary:
Our mission at Meta Open Source is to empower communities through open source, and we believe that it means building a welcoming and safe environment for all. As a part of this work, we are adding this banner in support for Ukraine during this crisis.

* add an ignore file

* [fix] FSDP: handle the lazy_init better

- when state_dict and load_state_dict is called, let'em not change
  the lazy_init state.

* changelog

* longer timeout

* Revert "longer timeout"

This reverts commit 00cc145fe86210a0972a1e7ba4f37531b9e091eb.

* testing

* adding the failed test

* fix the global to local id

* formatting

* more complete fix and test

* minor fix for an assert

* update changelog

* remove an extra line

* Update fairscale/nn/data_parallel/fsdp_optim_utils.py
Co-authored-by: anj-s <32556631+anj-s@users.noreply.github.com>

* Update fairscale/nn/data_parallel/fsdp_optim_utils.py
Co-authored-by: anj-s <32556631+anj-s@users.noreply.github.com>

* Update fairscale/nn/data_parallel/fsdp_optim_utils.py
Co-authored-by: anj-s <32556631+anj-s@users.noreply.github.com>

* addressed review comments
Co-authored-by: Min Xu <min.xu.public@gmail.com>
Co-authored-by: anj-s <32556631+anj-s@users.noreply.github.com>

* [FSDP] Add an arg for FSDP __init__

Add an arg, disable_reshard_on_root, for FSDP __init__ to handle the following issue
https://github.com/facebookresearch/fairscale/issues/878


For some cases (models wrapped by autowrap), the parameters (of root modules) needs to be sharded, and reshard_after_forward should not be set to False.
"disable_reshard_on_root" is for users to choose whether to force reshard_after_forward of root modules to be False or not.

* Update fully_sharded_data_parallel.py

Modified the description of the feature to explain more clear.

* Update fairscale/nn/data_parallel/fully_sharded_data_parallel.py

Update the comments for disable_reshard_on_root
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* Modified the comments

Modified the comments of disable_reshard_on_root

* Add a new argument for OSS __init__

Add a new argument for OSS __init__ to force the OSS to apply "_broadcast_object" for rebuilding the sharded optimizer. For more details, please see https://github.com/facebookresearch/fairscale/issues/937



* Remove redundant space

Remove redundant space
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* checkpoint tests

* checkpoint tests

* fix tests

* lint fixes

* remove prints

* lint fixes

* add comments

* add changelog

* more cleanup

* lint fix

* add benchmarks for fsdp

* fix lint errors

* clean up

* clean up unused flags

* add the benchmarks

* remove unused args

* fix lint errors

* fix lint errors

* update command line

* add support for multiple devices

* try full fp16 mode

* try full fp16 mode

* lint errors

* merge main

* lint errors

* lint errors

* lint error

* update intersphinx mapping for numpy

* update intersphinx mapping for numpy

* skip test

* added golden configs

* use synthetic benchmarks

* fix fn name

* fix cuda device id

* fix verify

* lint fix

* Update CHANGELOG.md

Adding https://github.com/facebookresearch/fairscale/pull/930

 to changelog

* Update CHANGELOG.md
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* [fix] Add option to wrap root module in auto_wrap

* Fix unit-test comment

* adding a few more tests to make expected behavior clear

* move changes to wrap policy as suggested

* set default to false

* revert pre-commit change

* revert pre-commit change 2
Co-authored-by: Ruan Silva <ruanrms@fb.com>

* update pytest versions

* [test] test related changes

- upgrade to newer pytorch versions
- added function to make test more deterministic on A100 and TF32
- fixed some tests so that they are correctly skipped on a single GPU system

* more fixes

* formatting overly long lines

* format

* better test without trigger a warning

* fix an optim state bug with newer pytorch

- adam optimizer seems to return "step" as a singleton tensor now in the
nightly build
- this fixes it assumeing non-tensor value can still be loaded back by
the optimizer

* improve oss.py

- use min_loss for regression checking is a bit more reliable
- also increased the num epochs from 10 to 12

* small oss.py fix

* Update fairscale/nn/data_parallel/fully_sharded_data_parallel.py
Co-authored-by: Min Xu <min.xu.public@gmail.com>

* skipping one more test

* formatting

* minor fix and copyright header

* comment
Co-authored-by: Min Xu <min.xu.public@gmail.com>

* [FSDP] Add an arg for FSDP __init__

Add an arg, disable_reshard_on_root, for FSDP __init__ to handle the following issue
https://github.com/facebookresearch/fairscale/issues/878


For some cases (models wrapped by autowrap), the parameters (of root modules) needs to be sharded, and reshard_after_forward should not be set to False.
"disable_reshard_on_root" is for users to choose whether to force reshard_after_forward of root modules to be False or not.

* Update fully_sharded_data_parallel.py

Modified the description of the feature to explain more clear.

* Update fairscale/nn/data_parallel/fully_sharded_data_parallel.py

Update the comments for disable_reshard_on_root
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* Modified the comments

Modified the comments of disable_reshard_on_root
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* update intersphinx mapping for numpy

* update intersphinx mapping for numpy

* update pytorch mapping and disable test

* [feat] add CosFace paper's LMCL to MEVO

- added baseline algorithm to the reference kernel
- added MEVO version of LMCL
- added unit test to verify it is correct with respect to the reference as well as its memory usage

* updated changelog
Co-authored-by: Min Xu <min.xu.public@gmail.com>

* [minor] better assert in backward

* mypy
Co-authored-by: Min Xu <min.xu.public@gmail.com>

* [fix] reduce unit test memory

* set seed in CI

* fix random seed function

* giving up CI, //sigh

* Add FairScale FSDP adoptions logging

* Add FairScale FSDP adoptions logging

* release 0.4.5

* added some content for the release

* fixed a format issue.

* [skip ci] first commit

* [skip ci] gradient scaler example

* [skip ci] adding feed forward toy example

* [skip ci] adding types

* [skip ci] adding backward hook

* [skip ci] update

* [skip ci] working feed forward example

* [skip ci] working feed forward example

* [skip ci] use named_modules instead of named_children

* [skip ci] adding new file

* [skip ci] clean up

* [skip ci] implement unscale function

* [skip ci] implement unscale function

* [skip ci] removing old file

* [skip ci] removing some more old files

* [skip ci] making unscale function generic

* [skip ci] adding test for vision model

* [skip ci] adding identity layer

* [skip ci] cleanup files

* [skip ci] refactoring

* [skip ci] more refactoring

* [skip ci] added functionality to update scale

* [skip ci] data loader clean up

* [skip ci] implemented inf checks and update scale functions

* [skip ci]code clean up. added...

* fixed padding size of input tensor for reduce scatter, and fixed an error that assigned wrong group

* Update fairscale/nn/data_parallel/fully_sharded_data_parallel.py
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

* added changelog

* fixed some commit.

* added unit test to ensure the reduce_scatter process group size is correct in default cases. And fall back to default process grouop when the reduce_scatter process group has the wrong size.

* throw an error instead of rolling back to use default process group for reduce_scatter_process_group

* Revert "throw an error instead of rolling back to use default process group for reduce_scatter_process_group"

This reverts commit eab5620da3b726ea55d3088ae4ca10d94dcdf4d9.

* added check for None to avoid unit test failure

* fixed an error to avoid the unit tests failure
Co-authored-by: Min Xu <24926999+min-xu-ai@users.noreply.github.com>

[chore] Update the CHANGELOG to add details about the new feature that enables reduce_scatter overlap in backward propagation (#906)

* updated the change log

* improve the change log

* enable reduce scatter overlap with other operations

* fixed unit tests and added docstrings for the new parameters for fsdp

* fixed more unit tests

* fixed unit tests

* avoided the pickle error on process_group_reduce_scatter

* removed an unnecessary parameter in unit tests

* remove unnecessary prints

* fixed the docstring

* skipped the test_offload unit test because this unit test failed in the main branch

* removed the enable_reduce_scatter_overlap API parameter

* added doc string for the defualt value of process_group_reduce_scatter parameter

* fixed a syntax bug

* fixed a bug which cause unitest failure

* removed the all_gather in the ProcessGroupName enum

* added more comment

* changed the default value of process_group_reduce_scatter from None to ProcessGroupName.reduce_scatter

* FullyShardedDataParallel: only return full state dict on rank 0

* Add flag and make rank 0 only optional

* Add tests

* Add docs

* address comments

* update comments

* update torch nightly version

* update torchvision number for torch nightly dependence

* add changelog

* Update CHANGELOG.md

* Update CHANGELOG.md