# How to start fsdp2 when using trainer?

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-start-fsdp2-when-using-trainer%2F151885%3E%5C
**Category:** 🤗Transformers\
**Created:** [April 23, 2025, 5:01am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-start-fsdp2-when-using-trainer%2F151885 "2025-04-23T05:01:24Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Tavish9](https://avatars.discourse-cdn.com/v4/letter/t/ccd318/32.png) [@Tavish9](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTavish9)%5C
**Post date:** [April 23, 2025, 5:01am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhow-to-start-fsdp2-when-using-trainer%2F151885%2F1 "2025-04-23T05:01:24Z")

</div>

Are there some ways to start fsdp2 using fsdp config or CLI arguments?

Currently, `--fsdp` would initialize `FullyShardedDataParallelPlugin(fsdp_version=1)`.

Trainer initializes Accelerator using the self-generated args, and we are now allowed to pass a FullyShardedDataParallelPlugin to Accelerator.

Any ideas?