GitHub · Where software is built

Labels Milestones New issue

Support passing in nn.Parameter to a helion kernel

PyTorch interop

#1356

· Chillee opened

on Jan 27, 2026

helion default config hangs when indexing tensor with enough dims.

#1354

· Chillee opened

on Jan 27, 2026

[RFC] Helion kernel + Inductor ops prologue / epilogue fusion

PyTorch interop

#1346

· yf225 opened

on Jan 23, 2026

Helion bf16_gemm_gelu(erf): failing for accuracy on MI300x and RX 9060XT for small matrices

#1342

· LironKesem opened

on Jan 20, 2026

Incorrect tiling with jagged output?

#1339

· d-laub opened

on Jan 19, 2026

Autotuning with tensors >= 2^32 errors when block pointers are used (and why isn't <code>HELION_AUTOTUNE_IGNORE_ERRORS=1</code> on by default?)

#1335

· Chillee opened

on Jan 13, 2026

While doing accuracy checks, run with determinism mode enabled (i.e. map <code>torch.empty</code> to be all nans or something)

#1334

· Chillee opened

on Jan 13, 2026

Enable User-Driven Early Termination and Caching in Autotuning

#1331

· rafaykhurram opened

on Jan 9, 2026

Enhance System Reliability with Checkpoint-Based Resumability

#1330

· rafaykhurram opened

on Jan 9, 2026

Serving Helion Models

#1326

· pramodith opened

on Jan 7, 2026

[RFC] Support runtime extension of kernel configs via environment variables

#1296

· xiaohongchen1991 opened

on Dec 22, 2025

[RFC] Structured utilities for specialization key construction

#1292

· xiaohongchen1991 opened

on Dec 19, 2025