●Change8

Migrating to Accelerate v1.8.0

Version v1.8.0 introduces 1 breaking change. This guide details how to update your code.

Released: 6/19/2025

1
Breaking Changes
4
Migration Steps
12
Affected Symbols

⚠️ Check Your Code

If you use any of these symbols, you need to read this guide:

FSDPv2ipex.optimizetorch.nn.Module.compiletorch.compileCCL_WORKER_COUNTKMP parametersDeepSpeed engineSwanLabaccelerator().load_state()dtype_byte_sizete.LayerNormTorchTensorParallelPlugin

Breaking Changes

●Issue #1

FSDPv2 setup is now more restrictive due to simplification of composition with features like FP8, torch.compile, and activation checkpointing. Users must adapt their FSDPv2 setup to the new, simplified composition method to avoid errors and ensure performance.

Migration Steps

  1. 1
    Review and update FSDPv2 setup to align with the new, simplified composition rules.
  2. 2
    For users relying on `ipex.optimize`, transition to native PyTorch optimizations if using PyTorch 2.8 or newer.
  3. 3
    Users training on Intel CPUs should verify or update their environment variables related to `CCL_WORKER_COUNT` and `KMP` parameters for performance gains.
  4. 4
    If encountering issues with tracker initialization in distributed setups, note that initialization is now deferred.

Release Summary

This release introduces major refactoring for FSDPv2 setup, adding FP8 support, and significantly enhancing performance and stability for Intel CPU/XPU users. It also deprecates `ipex.optimize` and integrates SwanLab as a new experiment tracker.

Need More Details?

View the full release notes and all changes for Accelerate v1.8.0.

View Full Changelog