Search results

Add Support for MiCS Shards on Deepspeed

Status: Open.

#3442 In huggingface/accelerate;

· sam-h-bean opened on Mar 13, 2025

Issue with distribution strategy for sagemaker

Status: Open.

#3440 In huggingface/accelerate;

· felix-dumont-neoxia opened on Mar 13, 2025

ppc64le version is significantly outdated.

Status: Open.

#3436 In huggingface/accelerate;

· yaniker opened on Mar 11, 2025

Using Deepspeed zero 3 to save Checkpoint

Status: Open.

#3435 In huggingface/accelerate;

· JYX1216 opened on Mar 11, 2025

Issue with model generations and using GRPOTrainer with device_map='auto' passed while loading model.

Status: Open.

#3434 In huggingface/accelerate;

· debdeepsanyal opened on Mar 10, 2025

Could not load random states in Pytorch >= 2.4

Status: Open.

#3433 In huggingface/accelerate;

· XiaoyuBIE1994 opened on Mar 10, 2025

Slowdown caused by accelerate or pytorch?

Status: Open.

#3431 In huggingface/accelerate;

· ZeshengLiu22 opened on Mar 9, 2025

Can i run a big model on several machines ?

Status: Open.

#3429 In huggingface/accelerate;

· shaofengzeng opened on Mar 8, 2025

Problem with accelerate == 1.4.0 and deepspeed ==0.16.4 when training NLP models

Status: Open.

#3428 In huggingface/accelerate;

· CaoYiwei opened on Mar 8, 2025

Multi-Node Training Fails with NCCL Communication Errors on NVIDIA DGX Cloud

Status: Open.

#3426 In huggingface/accelerate;

· mahdip72 opened on Mar 7, 2025

Unwrapping for generation with FSDP

Status: Open.

#3425 In huggingface/accelerate;

· VityaVitalich opened on Mar 6, 2025

Partial accelerator.prepare() Usage

Status: Open.

#3422 In huggingface/accelerate;

· zhanglixuan0720 opened on Mar 6, 2025