Search results

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

Status: Open.

#1334 In pytorch/helion;

· Chillee opened on Jan 13, 2026

Enable User-Driven Early Termination and Caching in Autotuning

autotuning

Status: Open.

#1331 In pytorch/helion;

· rafaykhurram opened on Jan 9, 2026

Enhance System Reliability with Checkpoint-Based Resumability

autotuning

Status: Open.

#1330 In pytorch/helion;

· rafaykhurram opened on Jan 9, 2026

Serving Helion Models

Status: Open.

#1326 In pytorch/helion;

· pramodith opened on Jan 7, 2026

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

deployment

Status: Open.

#1296 In pytorch/helion;

· xiaohongchen1991 opened on Dec 22, 2025

[RFC] Structured utilities for specialization key construction

specialization

Status: Open.

#1292 In pytorch/helion;

· xiaohongchen1991 opened on Dec 19, 2025

[RFC] automatic loop separation to optimize launch grid by the autotuner

autotuning

Status: Open.

#1286 In pytorch/helion;

· bringlein opened on Dec 18, 2025

Tips for faster but still performant autotuning?

autotuning

Status: Open.

#1274 In pytorch/helion;

· tyler-romero opened on Dec 16, 2025

Support triton.heuristics in hl.triton_kernel()

language support

Status: Open.

#1265 In pytorch/helion;

· oulgen opened on Dec 13, 2025

Support triton.autotune in hl.triton_kernel()

language support

Status: Open.

#1264 In pytorch/helion;

· oulgen opened on Dec 13, 2025

Windows support

windows

Status: Open.

#1245 In pytorch/helion;

· cooldome opened on Dec 10, 2025

[RFC] Allow user override of configurations at runtime (or make register of block sizes conditionally)

deployment

specialization

Status: Open.

#1242 In pytorch/helion;

· bringlein opened on Dec 10, 2025