Pull Requests huggingface/accelerate

chore(deps): bump the actions group across 1 directory with 7 updates dependencies github_actions

#4148 opened 2026-08-03 07:59 by dependabot[bot]

Fix convert_model_to_fp8_ao converting the first and last linear layers

#4147 opened 2026-08-02 16:45 by vineethsaivs

Fix moved nd_parallel.py example path in context-parallelism guide

#4146 opened 2026-07-31 16:02 by latent-9

[FSDP2] Add additional_modules_to_shard for sharding custom modules

#4145 opened 2026-07-31 06:45 by yuxinyuan

[docs] Add a concept guide on sharding iterable datasets

#4144 opened 2026-07-30 20:48 by zokost

Add end-to-end distributed test suite (Phase 0: DDP smoke + env parity)

#4143 opened 2026-07-30 11:21 by Titus-von-Koeller

[FSDP2] Shard eagerly initialized optimizer state

#4142 opened 2026-07-29 13:00 by sdjasj

[TP] Keep optimizer state aligned after parameter swaps

#4141 opened 2026-07-29 12:50 by sdjasj

[FSDP] Validate activation checkpointing auto-wrap policy

#4140 opened 2026-07-29 12:45 by sdjasj

[launch] Report one actionable diagnostic when distributed ranks fail (and make --quiet work for multi-GPU)

#4139 opened 2026-07-28 18:32 by mosafariuk

Reject sharded-checkpoint shard paths that escape the checkpoint folder

#4138 opened 2026-07-25 17:05 by AbdullahRasheed45

Document bnb_quantization_config in replace_with_bnb_layers

#4136 opened 2026-07-24 05:08 by LuShadowX

Fix typos and a broken kwarg in the sequence parallelism guide

#4135 opened 2026-07-24 04:40 by Savi-Swar

[Examples] Add distributed image captioning inference example script (#3078)

#4133 opened 2026-07-21 09:11 by Rajeev91691

Reach through the DeepSpeed optimizer wrapper in AcceleratedOptimizer.eval()

#4130 opened 2026-07-20 21:01 by vineethsaivs

Fix BatchSamplerShard.__len__ overcounting when split_batches=True and even_batches=False

#4127 opened 2026-07-18 13:39 by pepepeboom

Add level and msg type hints to MultiProcessAdapter.log

#4125 opened 2026-07-18 01:55 by RudrenduPaul

Add module: nn.Module type hints to ModelHook lifecycle methods

#4124 opened 2026-07-18 01:55 by RudrenduPaul

Add missing -> bool return type hints in utils/imports.py

#4123 opened 2026-07-17 05:57 by RudrenduPaul

Fix write_basic_config clobbering MLU config to CPU

#4121 opened 2026-07-16 21:13 by Osamaali313

Fix concatenate() for non-tensor batch fields across multiple batches

#4120 opened 2026-07-16 08:07 by Kaif10

Fix IterableDatasetShard.__len__ to account for split_batches

#4118 opened 2026-07-16 03:47 by chuenchen309

docs: clarify multi-process non-tensor batch limitation

#4117 opened 2026-07-14 20:37 by DebadityaHait

Fix environment conversion docstring example

#4115 opened 2026-07-13 23:10 by cupkk

Fix typos in comments, docs, and examples

#4113 opened 2026-07-09 20:58 by maxtaran2010

docs: add community health files and README updates

#4112 opened 2026-07-09 12:04 by Mukller

Add per-module fp32 precision for norms under FSDP2

#4110 opened 2026-07-08 15:59 by winglian

Expose force_recompute_fp8_weight_in_bwd on AORecipeKwargs

#4109 opened 2026-07-08 15:14 by winglian

Fix sequence parallel bool parsing

#4108 opened 2026-07-08 06:27 by lxingy3

Fix megatron_lm bool launch flags parsing "False" as True

#4107 opened 2026-07-08 00:51 by teddytennant