deps-py(deps): bump accelerate from 0.34.2 to 1.13.0 in /apps

[dependabot]

Bumps accelerate from 0.34.2 to 1.13.0.

Release notes

Sourced from accelerate's releases.

v1.13.0: Neuron support, IPEX removal, and distributed training fixes

AWS Neuron support

We now have support for AWS Neuron (Trainium/Inferentia) devices. Thanks @​michaelbenayoun for adding this.

• Neuron integration by @​michaelbenayoun in huggingface/accelerate#3935

XPU Improvements

We've removed IPEX dependency and improved device-agnostic code for XPU.

• using spawn instead of fork for XPU device by @​kaixuanliu in huggingface/accelerate#3884
• Remove ipex by @​yao-matrix in huggingface/accelerate#3883
• enhance new codes to XPU, and make them be device agnostic by @​yao-matrix in huggingface/accelerate#3890
• Fix KMP_AFFINITY incorrectly set for non-CPU training by @​hexfaker in huggingface/accelerate#3912

FSDP2 Improvements

We've added a bunch of important fixes for FSDP2 users: upcasting only grad-requiring params, better tied embedding errors, DCP optimizer loading, bf16 optimizer step crash fix, and torch < 2.7.0 compatibility.

• Upcast FSDP2 parameters only if requires_grad by @​ojh31 in huggingface/accelerate#3848
• Fix FSDP2 tied embedding errors with targeted ValueError guidance by @​amanzoni1 in huggingface/accelerate#3878
• bug: fsdp cannot load optimizer state using dcp by @​flymin in huggingface/accelerate#3904
• fix crash in optimizer.step when fsdp2 is enabled and model is bfloat16 by @​sywangyi in huggingface/accelerate#3905
• Fix FSDP2 crash with ignored_params on torch < 2.7.0 by @​Mr-Neutr0n in huggingface/accelerate#3924

DeepSpeed Sequence Parallelism

We've added several fixes to the DeepSpeed + Sequence Parallelism integration introduced in v1.12.0, including evaluation support during SP training and proper process group handling.

• [SP] fix loss computation example by @​kashif in huggingface/accelerate#3858
• [SP and CP] error out if both CP and SP enabled by @​kashif in huggingface/accelerate#3862
• DeepSpeed has its own process group by @​kashif in huggingface/accelerate#3916
• [Deepspeed] skip device mesh creation when deepspeed and sp_size >1 by @​kashif in huggingface/accelerate#3914
• Enable evaluation during deepspeed Sequence Parallel by @​jp1924 in huggingface/accelerate#3917

FP8

We've enhanced FP8 training. Thanks @​shimizust for fixing torchao support.

• Fix FP8 torchao default config with padding and FSDP2 all-gather support by @​shimizust in huggingface/accelerate#3831
• Fix execution with Transformer Engine by @​ksivaman in huggingface/accelerate#3852
• add MS-AMP deprecation warnings by @​neha222222 in huggingface/accelerate#3857

Performance

Accelerate now imports faster by deferring heavy dependencies, and torch.compile hooks are disabled lazily.

• Faster import by @​SunMarc in huggingface/accelerate#3953
• lazy compile disable by @​SunMarc in huggingface/accelerate#3947
• Disable hook compile by @​SunMarc in huggingface/accelerate#3888

Minor fixes

• Allow non-Tensor values in a batch with dispatch_batches=True by @​tomaarsen in huggingface/accelerate#3850
• fix module and optimizer parameter mismatch before prepare_tp_ by @​naomili0924 in huggingface/accelerate#3845
• Fix KeyError in extract_model_from_parallel for partial torch.compile by @​amanzoni1 in huggingface/accelerate#3881
• Fix hf_device_map device index comparison in prepare_model by @​rezaqorbani in huggingface/accelerate#3895
• Fix StatefulDataLoader KeyError with num_workers > 0 by @​veeceey in huggingface/accelerate#3931
• Fix stateful dataloader DDP by @​SunMarc in huggingface/accelerate#3952
• Fix: Remove duplicate W&B initialization in offline mode by @​shantanugupta2004 in huggingface/accelerate#3886

... (truncated)

Commits

• e6ee133 Release: v1.13.0
• 2a7e27f Fix testing ci (#3955)
• 0990ded Faster import (#3953)
• 5cf9cf8 fix-stateful-dataloader (#3952)
• beed693 Prepare TP fix (#3945)
• 8067aba Fix StatefulDataLoader KeyError with num_workers > 0 (#3931)
• 8ec83c8 Fix FSDP2 crash with ignored_params on torch < 2.7.0 (#3924)
• 7554afb Fix mutable default in Megatron init and IndexError on empty ModuleList (#3944)
• 23f2ab3 Fix logging logic when in_order is set to True (#3280)
• 58c3605 docs: update low-precision training docs to reflect MS-AMP deprecation (#3929)
• Additional commits viewable in compare view

[dependabot]

Labels

The following labels could not be found: supply-chain, w763. Please create them before Dependabot can add them to a pull request.

Please fix the above issues or remove invalid values from dependabot.yml.