Sync with Microsoft ONNX Runtime - 20062026#1154

Open

ai-fw-intg wants to merge 9 commits into

ovep-developfrom

sync_msft_20062026

Commits on Jun 18, 2026

Harden QMoE against integer overflow and partial K tiles (#29067)
tianleiwu
authored
[CUDA] Add decode-optimized LinearAttention (GatedDeltaNet) kernels (#28985)

tianleiwu
and
Copilot
authored

Commits on Jun 19, 2026

Fix Attention is_causal bottom-right alignment for external KV cache (onnx#8068, #28904) (#28958)

titaiwangms
and
Copilot
authored
Bump ws in /js/react_native/e2e (#29057)
dependabot[bot]
authored
[CUDA] Optimize FlashDecode split planning for local-window GQA (#29161)
tianleiwu
authored
[CUDA] Remove TensorRT fused causal attention kernels (#29143)
tianleiwu
authored
[MLAS] KleidiAI fix igemm regression (#28571)

authored
Fixed failing KleidiAI NHWC unit tests (#29010)
martin-klacer-arm
authored
Merge remote-tracking branch 'origin/master' into sync_msft_20062026
AIFrameworksIntegration
committed