DeepSpeed adaption for flux-kontext #12240
Conversation
|
Cc: @sayakpaul @a-r-r-o-w Please take a look at this PR. Thanks! |
leisuzz
force-pushed
the
kontext
branch
3 times, most recently
from
3950d2e to
3094c2b
Compare
|
@sayakpaul I've also added ds support for load/save ckpt function. Thanks! |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
|
@sayakpaul Please approve the workflow, thx :) |
|
Hi @sayakpaul, It looks weird but I think the original code does not need the empty line. I added the empty line to it. Thanks :) |
|
Cc: @sayakpaul, I checked with ruff and make quality, they all passed. Sorry for the trouble and thanks for your help :) |
|
Cc: @sayakpaul @a-r-r-o-w |
|
@leisuzz could I request you to not ping the maintainers multiple times on a PR? This has happened in the past, so, I thought I would bring it up. |
|
@sayakpaul Sorry about that |
Hi @sayakpaul, |
|
Hi @sayakpaul, I think the failing tests are unrelated. Thank you for your help :) |
What does this PR do?
"ValueError: At least one of the dataloaders passed to accelerate.prepare() has None as batch size. Please set an integer value in train_micro_batch_size_per_gpu in the deepspeed config file or assign integer value to AcceleratorState().deepspeed_plugin.deepspeed_config['train_micro_batch_size_per_gpu']."I think the current batch_sampler does not support deepspeed. So I added this function to support it.
Before submitting
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.