Repository navigation

[DSv4][P1-5] Implement CUDA and Triton rmsnorm_residual - #414

Open

why1te wants to merge 4 commits into

RL-Align:dsv4-p1-devfrom

why1te:dsv4-p1-dev

Open

[DSv4][P1-5] Implement CUDA and Triton rmsnorm_residual#414
why1te wants to merge 4 commits into
RL-Align:dsv4-p1-devfrom
why1te:dsv4-p1-dev

Commits on Sep 16, 2026
- [DSv4][P1-5] Implement CUDA and Triton rmsnorm_residual
  why1te
  committed
  9ff7326
- [DSv4][P1-5] Align FP32 gamma inputs and refresh golden hashes
  why1te
  committed
  3abb6d6
- [DSv4][P1-5] Document rmsnorm_residual operator
  why1te
  committed
  b63452c
Commits on Oct 6, 2026
- Revert unrelated formatting changes
  zhangj1an
  committed
  95ec4e5