GitLab

* typo, sorry about that

* small perf fix

* Better handling of the callback queue, try to consume it as we go.

* dumping buckets for the reduce part, always the same unused params issue

* lintfixes

* come on black

* Update tutorial_pipe_multiprocess.py

make RANK global like the other tutorials
Co-authored-by: Vittorio Caggiano <caggiano@gmail.com>

* Update README.md

* Update README.md

update capitalization
Co-authored-by: Vittorio Caggiano <caggiano@gmail.com>

* [feat] add CPU support to tutorials in examples

* now works on a machine without cuda
* fixes some minor typos

* [cleanup] factorize tutorials in examples

* collects duplicate code across tutorials in helpers.py

* [fix] getData in tutorials now returns iterable

* [doc]: AdaScale example and notes

* formatted notes correctly as suggested by Benjamin

* added feature and unit test to make sure lr_scheduler works

* update the example with lr_scheduler

* fixed doc with "make html"

* addressed Mike's suggestions

* better ddp adascale tests

* make sure the single node test use the same test cases and expected gains

* added unit test that covers smoothing factor

- tested by re-introducing the bug and see the test fail as expected.

* [doc] updating the pipe balance doc a bit

- Also added a warning to pipeline.py when the partition output is not
supported.

* addressed Mandeep's comment

* unit test checking ddp and sharded_ddp equivalence, reproducing the issue that Sean spotted
* fixing the issue, not counting requests in flight properly
* adding a multiple optimizers case

* removing strict typing requirement, broken by ClassyVision

Thanks Jessica for the heads up !

* add logo

* Update README.md
Co-authored-by: Vittorio Caggiano <caggiano@fb.com>

* proper unit testing, but no other solution than disabling bucketing for now, couple of options tested do not work

* added AdaScale to README

* [adascale] added gradient accumulation

- added gradient accumulation
- tested with cifar full trainings with different value of accumulation
and verified the full accuracy is obtained
- also removed the patch optimize flag until we need it

* [adascale] adding pytest

- added basic and ddp tests and grad_accum
- closes #195

* added changelog

* added ddp grad_accum test

* moved ddp and non-ddp tests into separate files

* added checkpoint test

* more doc

* addressed Mike's comments

Fixes #190

* fallback on internal pytorch numbering

Fixing the relative positions of the html docs.

* make the basic example usable out of the box
* clarify

* testing median and MAD

* synchronize on kernels to make sure that we're measuring the actual completion time

* adjusting the circleci threshold, not that the speed has regressed but because we measure proper cuda execution time

* rewrite using autograd and Variable execution queue to make the reduce automatic
* share buckets with OSS to remove duplication
* some speed still likely on the table since the speed vs. bucketing does not match expectations, could be a follow up

Fixes #190

* reverting a change which slipped in #188

* Add CPU support for pipe.py benchmarks, CUDA-free

* adding a shard-aware GradScaler wrap, credits to Sean Naren for the idea
* adding stubs & explanations in the documentation

- removed experimental warning as we have validated it on cifar and
imagenet, transformer is looking good so far too.
- fixed API doc formatting
- make it consistent with the other code in the repo
- tested by making the doc locally and inspect the results

add a clip gradients util, equivalent to torch's but aware of the sharded states. Add a corresponding unit test