unslothai/unsloth

[Bug] Multi-gpu finetuning in Unsloth

Open

#3,248 opened on Sep 1, 2025

View on GitHub
 (2 comments) (0 reactions) (0 assignees)Python (64,271 stars) (5,658 forks)batch import
help wantedmultigpu

Description

  1. Did you update? pip install --upgrade unsloth unsloth_zoo -->yes
  2. Colab or Kaggle or local / cloud -->cloud
  3. Number GPUs used, use nvidia-smi --?> 4
  4. Which notebook? Please link!
  5. Which Unsloth version, TRL version, transformers version, PyTorch version?
  6. Which trainer? SFTTrainer, GRPOTrainer etc --> SFTTrainer
Put Minimal code to reproduce error here ###Remove Hugging Face token###

🦥 You can also ask via our Reddit page: https://www.reddit.com/r/unsloth/

I am able to finetuned Gemma 3 4B model using unsloth in one gpu it is working fine, but when I am trying to use it in multi-gpu settings, I am not able to use it properly. Any help regarding multi-gpu support will be appriciate. Please help to set up multi-gpu finetuning through unsloth. Thanks

Contributor guide

Tech stack
pythonpytorch
Domain
machine learningai
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
needs investigation
Prerequisites
Multi GPU basicsPyTorch DDP
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
Investigate Unsloth's current multi GPU support by reviewing the official documentation and examples. Check the repository for any existing issues or PRs related to multi GPU training. Look into the use of PyTorch's DistributedDataParallel or Fully Sharded Data Parallel (FSDP) within Unsloth. Consider adding a minimal example of multi GPU finetuning with SFTTrainer and Gemma 3 4B.