# Tensor parallelism inference

**URL:** <https://discuss.huggingface.co/t/tensor-parallelism-inference/98071>\
**Category:** 🤗Transformers\
**Created:** [July 18, 2024, 5:47am UTC](https://discuss.huggingface.co/t/tensor-parallelism-inference/98071 "2024-07-18T05:47:25Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![yumj](https://avatars.discourse-cdn.com/v4/letter/y/8491ac/32.png) [@yumj](https://discuss.huggingface.co/u/yumj)\
**Post date:** [July 18, 2024, 5:47am UTC](https://discuss.huggingface.co/t/tensor-parallelism-inference/98071/1 "2024-07-18T05:47:25Z")

</div>

When i want to use tensor parallelism during the model inference , I find the parallelism is supported on training. How to customize tensor parallelism？