# Helion

**URL:** https://dev-discuss.pytorch.org/c/helion/31.md

[Latest](https://dev-discuss.pytorch.org/latest.md) · [Categories](https://dev-discuss.pytorch.org/categories.md)

---

## [About the Helion category](https://dev-discuss.pytorch.org/t/about-the-helion-category/3315)

<div class="topic-metadata">

**Author:** [@albanD](https://dev-discuss.pytorch.org/u/albanD)\
**Replies:** 0\
**Last updated:** [February 18, 2026, 5:50pm UTC](https://dev-discuss.pytorch.org/t/about-the-helion-category/3315 "2026-02-18T17:50:16Z")

</div>

Helion is a PyTorch-native domain-specific language (DSL) for authoring performance portable machine learning kernels. Helion is a PyTorch-native domain-specific language (DSL) for authoring high-performance machine lea…

---

## [We're working on Helion on TPU!](https://dev-discuss.pytorch.org/t/were-working-on-helion-on-tpu/3409)

<div class="topic-metadata">

**Author:** [@jongsok.choi](https://dev-discuss.pytorch.org/u/jongsok.choi)\
**Replies:** 0\
**Last updated:** [June 28, 2026, 2:24am UTC](https://dev-discuss.pytorch.org/t/were-working-on-helion-on-tpu/3409 "2026-06-28T02:24:07Z")

</div>

At this week at Google AI Systems DevLabs, we presented Helion on TPU for the first time publicly. This is a joint collaboration between Meta and Google. The talk was recorded and should be available on YouTube shortly! …

---

## [Helion Tutorial at PLDI conference!](https://dev-discuss.pytorch.org/t/helion-tutorial-at-pldi-conference/3408)

<div class="topic-metadata">

**Author:** [@jongsok.choi](https://dev-discuss.pytorch.org/u/jongsok.choi)\
**Replies:** 0\
**Last updated:** [June 28, 2026, 2:11am UTC](https://dev-discuss.pytorch.org/t/helion-tutorial-at-pldi-conference/3408 "2026-06-28T02:11:47Z")

</div>

We held a Helion tutorial at the PLDI conference (Writing Performance-Portable Kernels Simplified with Helion (PLDI 2026 - PLDI Tutorials) - PLDI 2026) introducing the language, its underlying motivation, overall design …

---

## [PyTorch blog on "From Minutes to Seconds: LLM-Guided Autotuning for Helion Kernels"](https://dev-discuss.pytorch.org/t/pytorch-blog-on-from-minutes-to-seconds-llm-guided-autotuning-for-helion-kernels/3407)

<div class="topic-metadata">

**Author:** [@jongsok.choi](https://dev-discuss.pytorch.org/u/jongsok.choi)\
**Replies:** 0\
**Last updated:** [June 28, 2026, 2:02am UTC](https://dev-discuss.pytorch.org/t/pytorch-blog-on-from-minutes-to-seconds-llm-guided-autotuning-for-helion-kernels/3407 "2026-06-28T02:02:16Z")

</div>

Abstract: Helion, PyTorch’s domain-specific language (DSL) for performance portable machine learning kernels, heavily relies on autotuning for performance. Currently Helion searches utilize the Likelihood-Free Bayesian…

---

## [PyTorch blog on "Portable vLLM Model Inference Kernels in Helion"](https://dev-discuss.pytorch.org/t/pytorch-blog-on-portable-vllm-model-inference-kernels-in-helion/3406)

<div class="topic-metadata">

**Author:** [@jongsok.choi](https://dev-discuss.pytorch.org/u/jongsok.choi)\
**Replies:** 0\
**Last updated:** [June 28, 2026, 2:00am UTC](https://dev-discuss.pytorch.org/t/pytorch-blog-on-portable-vllm-model-inference-kernels-in-helion/3406 "2026-06-28T02:00:41Z")

</div>

Abstract: Helion kernels were integrated into vLLM for FP8 inference using Qwen3 models and evaluated across NVIDIA H100 and B200 GPUs. The experiments show that Helion provides a productive PyTorch-native workflow for…