build(deps): update accelerate requirement from ~=0.20.3 to ~=0.22.0

[dependabot[bot]]

Updates the requirements on accelerate to permit the latest version.

Release notes

Sourced from accelerate's releases.

v0.22.0: Distributed operation framework, Gradient Accumulation enhancements, FSDP enhancements, and more!

Experimental distributed operations checking framework

A new framework has been introduced which can help catch timeout errors caused by distributed operations failing before they occur. As this adds a tiny bit of overhead, it is an opt-in scenario. Simply run your code with ACCELERATE_DEBUG_MODE="1" to enable this. Read more in the docs, introduced via huggingface/accelerate#1756

Accelerator.load_state can now load the most recent checkpoint automatically

If a ProjectConfiguration has been made, using accelerator.load_state() (without any arguments passed) can now automatically find and load the latest checkpoint used, introduced via huggingface/accelerate#1741

Multiple enhancements to gradient accumulation

In this release multiple new enhancements to distributed gradient accumulation have been added.

• accelerator.accumulate() now supports passing in multiple models introduced via huggingface/accelerate#1708
• A util has been introduced to perform multiple forwards, then multiple backwards, and finally sync the gradients only on the last .backward() via huggingface/accelerate#1726

FSDP Changes

• FSDP support has been added for NPU and XPU devices via huggingface/accelerate#1803 and huggingface/accelerate#1806
• A new method for supporting RAM-efficient loading of models with FSDP has been added via huggingface/accelerate#1777

DataLoader Changes

• Custom slice functions are now supported in the DataLoaderDispatcher added via huggingface/accelerate#1846

What's New?

• fix failing test on 8GPU by @​statelesshz in huggingface/accelerate#1724
• Better control over DDP's no_sync by @​NouamaneTazi in huggingface/accelerate#1726
• Get rid of calling get_scale() by patching the step method of optimizer. by @​yuxinyuan in huggingface/accelerate#1720
• fix the bug in npu by @​statelesshz in huggingface/accelerate#1728
• Adding a shape check for set_module_tensor_to_device. by @​Narsil in huggingface/accelerate#1731
• Fix errors when optimizer is not a Pytorch optimizer. by @​yuxinyuan in huggingface/accelerate#1733
• Make balanced memory able to work with non contiguous GPUs ids by @​thomwolf in huggingface/accelerate#1734
• Fixed typo in __repr__ of AlignDevicesHook by @​KacperWyrwal in huggingface/accelerate#1735
• Update docs by @​muellerzr in huggingface/accelerate#1736
• Fixed the bug that split dict incorrectly by @​yuangpeng in huggingface/accelerate#1742
• Let load_state automatically grab the latest save by @​muellerzr in huggingface/accelerate#1741
• fix KwargsHandler.to_kwargs not working with os.environ initialization in __post_init__ by @​CyCle1024 in huggingface/accelerate#1738
• fix typo by @​cauyxy in huggingface/accelerate#1747
• Check for misconfiguration of single node & single GPU by @​muellerzr in huggingface/accelerate#1746
• Remove unused constant by @​muellerzr in huggingface/accelerate#1749
• Rework new constant for operations by @​muellerzr in huggingface/accelerate#1748
• Expose autocast kwargs and simplify autocast wrapper by @​muellerzr in huggingface/accelerate#1740
• Fix FSDP related issues by @​pacman100 in huggingface/accelerate#1745
• FSDP enhancements and fixes by @​pacman100 in huggingface/accelerate#1753
• Fix check failure in Accelerator.save_state using multi-gpu by @​CyCle1024 in huggingface/accelerate#1760
• Fix error when max_memory argument is in unexpected order by @​ranchlai in huggingface/accelerate#1759
• Fix offload on disk when executing on CPU by @​sgugger in huggingface/accelerate#1762
• Change is_aim_available() function to not match aim >= 4.0.0 by @​alberttorosyan in huggingface/accelerate#1769

... (truncated)

Commits

• 6b3e559 Release: v0.22.0
• 16e5113 Improve big model inference docs (#1872)
• 3122a61 Include a note to the forums in the bug report (#1871)
• c8682ae support custom slice function in DataLoaderDispatcher (#1846)
• 0768905 remove casting to FP32 when saving state dict (#1868)
• d087be0 add env variable for init_on_device (#1852)
• 41caaa5 Update fsdp_with_peak_mem_tracking.py (#1856)
• 21d1273 fix dispatch (#1855)
• 3cf7dee Loading logic safetensors (#1853)
• 64c586f support for ram efficient loading of model with FSDP (#1777)
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot merge` will merge this PR after your CI passes on it - `@dependabot squash and merge` will squash and merge this PR after your CI passes on it - `@dependabot cancel merge` will cancel a previously requested merge and block automerging - `@dependabot reopen` will reopen this PR if it is closed - `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually - `@dependabot show ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

[peakji]

@dependabot ignore this patch version

[dependabot[bot]]

OK, I won't notify you about version 0.22.0 again, unless you re-open this PR.