microsoft/onnxruntime

Pull Requests Commits

[QNN EP] Fix segfault on PadNodeGroupSelector::Check when dq_nodes is empty (#28733)

grybouilli committed 18 days ago

Verified a08e17aa

Fix Shape→Gather→TopK regression: preserve rank-1 single-element index output in data propagation (#29084)

titaiwangms committed 18 days ago

Verified 20ce99fe

Fix DropQDQ not dropping Q/DQ around opset-22 MaxPool (#29201)

tairenpiao committed 18 days ago

Verified c69373c4

Fix CPU GQA NaN output for right-padded batched prompts with rotary embeddings (#29069)

Copilot committed 18 days ago

Verified 701d88ff

[CUDA] Enable CUDA GQA QK-Norm and XQA decode (#29186)

tianleiwu committed 19 days ago

Verified 1472c16e

[Pipeline] Fix Python packaging pipeline failures (#29200)

hariharans29 committed 19 days ago

Verified 8c62bee6

Fix GroupQueryAttention right-padded rotary prefill CUDA test (#29218)

tianleiwu committed 19 days ago

Verified 14a6c9e9

Experimental C++ API update (#29142)

edgchen1 committed 19 days ago

Verified 8c856fb9

fix bound check on webgpu/slice (#28704)

Guenther Schmuelling committed 19 days ago

Verified bc26b968

Bump @babel/core from 7.29.0 to 7.29.6 in /js/react_native/e2e (#29156)

dependabot[bot] committed 19 days ago

Verified a0aec633

[CUDA] update doc for cuda contrib op GroupQueryAttention (#29173)

tianleiwu committed 19 days ago

Verified 8c08e943

Remove the dynamic WGSL generator (duktape/Node) path (#29141)

danielsongmicrosoft committed 19 days ago

Verified 50f0314b

Address github issue 29071 (#29158)

yuslepukhin committed 19 days ago

Verified 9ad4d58c

Build with abseil 20250814 (NVCC) (#28586)

mc-nv committed 19 days ago

Verified 5b361106

[CUDA] Enable XQA by default for FP16/BF16 GQA (#29046)

namgyu-youn committed 20 days ago

Verified 6be94ded

webgpu: fix GQA batched right-padded prefill with do_rotary (#29002)

qjia7 committed 20 days ago

Verified 5ad25165

Add int8/uint8 CPU support for SpaceToDepth and int8 for DepthToSpace (#29154)

ArsalanShakil committed 20 days ago

Verified 737d2661

Fix TransposeOptimizer type error on zero-point-less DequantizeLinear (#28716) (#29192)

tairenpiao committed 21 days ago

Verified d3746361

Fix XNNPACK Clip/Relu fusion when the producer output has other consumers (#28814)

tairenpiao committed 21 days ago

Verified 86cbd205

Initialize large-head FMHA shared memory per kernel variant (#29140)

Kevin-Li-2025 committed 21 days ago

Verified efd584f4

Bump protobufjs from 7.6.0 to 7.6.3 in /js/node (#29090)

maoger committed 21 days ago

Verified 5a20ad7b

Fix Dawn parallel build race condition in wasm_Release / build-wasm CI (#29040)

Copilot committed 21 days ago

Verified b83fa1b3

[CUDA] Enable XQA decode for GroupQueryAttention with attention sink (#29162)

tianleiwu committed 22 days ago

Verified 3718f7dc

[CUDA] Default QMoE GEMV fp16 accumulation for fp16 activations (#29166)

tianleiwu committed 22 days ago

Verified bceab03c

Fixed failing KleidiAI NHWC unit tests (#29010)

martin-klacer-arm committed 22 days ago

Verified ad8e2582

[MLAS] KleidiAI fix igemm regression (#28571)

martin-klacer-arm committed 22 days ago

Verified bd0cb9a4

[CUDA] Remove TensorRT fused causal attention kernels (#29143)

tianleiwu committed 22 days ago

Verified 3b77775f

[CUDA] Optimize FlashDecode split planning for local-window GQA (#29161)

tianleiwu committed 22 days ago

Verified 89a23d8b

Bump ws in /js/react_native/e2e (#29057)

dependabot[bot] committed 22 days ago

Verified 88468c2b

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

titaiwangms committed 22 days ago

Verified 2db2660a

Newer Older