Search results

Use <code>assert x.size(0) == y.size(1)</code> to unify shape symbols

symbolic shape

Status: Open.

#1216 In pytorch/helion;

· yf225 opened on Dec 5, 2025

Given a BoundKernel, give me a hash that can be cheaply applied to new inputs to tell whether this BoundKernel is valid

Status: Open.

#1209 In pytorch/helion;

· Chillee opened on Dec 4, 2025

torch.rand_like shape causes compilation error

Status: Open.

#1208 In pytorch/helion;

· hrpan opened on Dec 4, 2025

iterating over device-side shape doesn't work if it's nested loops

language support

Status: Open.

#1202 In pytorch/helion;

· Chillee opened on Dec 4, 2025

Compile mode symbol broadcasting error doesn't repro in interpret mode

interpret modeHELION_INTERPRET=1 (ref mode) related issuesHELION_INTERPRET=1 (ref mode) related issues

symbolic shape

Status: Open.

#1200 In pytorch/helion;

· yf225 opened on Dec 4, 2025

Print out output file path when printing out code

Status: Open.

#1192 In pytorch/helion;

· Chillee opened on Dec 3, 2025

Helion autotuner benchmarking improvements

autotuning

Status: Open.

#1187 In pytorch/helion;

· yf225 opened on Dec 3, 2025

configuration save and load question

Status: Open.

#1185 In pytorch/helion;

· zjin-lcf opened on Dec 2, 2025

[RFC] A helion_aot_autotune wrapper that makes it easy for folks to convert a helion kernel into a "deployable" kernel

Status: Open.

#1161 In pytorch/helion;

· Chillee opened on Nov 22, 2025

hl.triton_kernel does not copy over global var or recursive kernel call

Status: Open.

#1152 In pytorch/helion;

· oulgen opened on Nov 20, 2025

Define clearer semantics for when a config can be re-used with different inputs.

Status: Open.

#1133 In pytorch/helion;

· Chillee opened on Nov 14, 2025

Autotune over epilogue subtiling

autotuning

Status: Open.

#1120 In pytorch/helion;

· Chillee opened on Nov 12, 2025