All issues

Search results

Forward Convolution Kernel

Status: Open.

#195 In NVIDIA/TileGym;

· m9ssah opened on Sep 4, 2026

attention_mask is discarded in fmha_interface_wrapper: is a varlen or padding-mask prefill kernel wanted?

Status: Open.

#191 In NVIDIA/TileGym;

· Aryan95614 opened on Aug 25, 2026

Would cuTile ports of Liger's chunked preference losses (DPO, ORPO, SimPO, KTO) be accepted into the liger suite?

Status: Open.

#190 In NVIDIA/TileGym;

· Aryan95614 opened on Aug 25, 2026

Why cutile's rmsnorm required full register

Status: Open.

#169 In NVIDIA/TileGym;

· MARD1NO opened on Jul 13, 2026

[E2E Model Support] Add TileGym kernel integration for LFM2 MoE

Status: Open.

#154 In NVIDIA/TileGym;

· iamanishx opened on Jun 12, 2026

Add Gemma 4 text-only inference support to TileGym transformers examples

Status: Open.

#130 In NVIDIA/TileGym;

· duoan opened on May 21, 2026

Training Examples?

Status: Open.

#104 In NVIDIA/TileGym;

· vgoklani opened on Apr 9, 2026

Potential numerical accuracy issue in the fused_moe implementation

Status: Open.

#23 In NVIDIA/TileGym;

· huanghua1994 opened on Dec 29, 2025