NVIDIA/recsys-examples v26.05

v26.05

Repository: NVIDIA/recsys-examples

Tag: v26.05

Published: 2026-06-16T07:22:44Z

Prerelease: no

Release notes:

What's Changed

Features & Enhancements

• Add GR inference example by @cb521 in https://github.com/NVIDIA/recsys-examples/pull/414
• benchmark: update HSTU benchmark fixes and docs by @JacoCheung in https://github.com/NVIDIA/recsys-examples/pull/411
• feat(hstu): enable HSTU + DynamicEmb E2E training on Blackwell (sm_100) by @JacoCheung in https://github.com/NVIDIA/recsys-examples/pull/399
• [Fea] beam search sm8x support by @z52527 in https://github.com/NVIDIA/recsys-examples/pull/407
• Segmented unique optimization by @jiashuy in https://github.com/NVIDIA/recsys-examples/pull/417
• Add an example for disabling contextual mask in training by @geoffreyQiu in https://github.com/NVIDIA/recsys-examples/pull/395
• Recsys-FlexKV cpu breakdown benchmark by @Clebrate in https://github.com/NVIDIA/recsys-examples/pull/416

Bug Fixes

• update doc by @shijieliu in https://github.com/NVIDIA/recsys-examples/pull/402
• Update beam search benchmark RESULTS.md by @z52527 in https://github.com/NVIDIA/recsys-examples/pull/406
• Refactor dynamicemb benchmark by @jiashuy in https://github.com/NVIDIA/recsys-examples/pull/396
• Refactor _print_memory_consume to report real per-tier value bytes by @jiashuy in https://github.com/NVIDIA/recsys-examples/pull/408
• Refactor/prefetch select cleanup by @jiashuy in https://github.com/NVIDIA/recsys-examples/pull/409
• [dynamicemb] Add spawn-based multi-worker DataLoader option to MovieL… by @jiashuy in https://github.com/NVIDIA/recsys-examples/pull/412
• fix offload_try_wait by @Clebrate in https://github.com/NVIDIA/recsys-examples/pull/415

New Contributors

• @cb521 made their first contribution in https://github.com/NVIDIA/recsys-examples/pull/414
• @Clebrate made their first contribution in https://github.com/NVIDIA/recsys-examples/pull/415

Full Changelog: https://github.com/NVIDIA/recsys-examples/compare/v26.04...v26.05