microsoft/onnxruntime

Pull Requests Commits

apsonawane committed 5 days ago

327e2dd2

Merge branch 'main' into asonawane/rotary

apsonawane committed 9 days ago

e95943c6

fix(ci): incorrect identity for azcopy (#29274)

sanaa-hamel-microsoft committed 9 days ago

Verified 5f49a372

Fix unbounded lifetime on WithOutputTensor in Rust bindings (#29251)

sayanshaw24 committed 9 days ago

Verified 0271287d

Address review feedback: drop stray blank line in mainline helper and add rank-3 no-num_heads test for hidden_size guard

apsonawane committed 9 days ago

71a33520

[CPU] Add FP32 GEMV decode kernel for GroupQueryAttention (#29216)

tianleiwu committed 9 days ago

Verified a203dfaf

webgpu: Enable FlashAttention for batched GQA with right-padded prompts (#29247)

qjia7 committed 10 days ago

Verified 92b4c663

Fix CPU Attention causal mask alignment (#29050)

FuZoe committed 11 days ago

Verified d4b01e43

[CUDA] Support user compute stream with CUDA graph in CUDA plugin EP (#29221)

tianleiwu committed 11 days ago

Verified 3b022ecb

Update cpuinfo to include cpuinfo_deinitialize(), fix QNN ETW logging, GQA underflow, and ep_weight_sharing_ctx_gen build (#28245)

crvineeth97 committed 11 days ago

Verified 4ed5a4ad

Add flash attention for non-quantized CPU GroupQueryAttention (#28962)

tianleiwu committed 11 days ago

Verified 996cea1e

Address review feedback: drop redundant mainline guards, revert contrib width-check tightening, tighten test assertions

apsonawane committed 11 days ago

89def7b0

Merge branch 'main' into asonawane/rotary

apsonawane committed 11 days ago

bf5813d2

Clamp derived sequence lengths and KV-cache index in CUDA GroupQueryAttention (#29240)

jiafatom committed 11 days ago

Verified eade9ea9

Relax CompileModel validation to accept zero-input OrtModel graphs (#28771)

adrastogi committed 11 days ago

Verified b77b3123

Fix Linux CUDA 13.3 build (abseil + CCCL parse errors) (#29042)

tianleiwu committed 11 days ago

Verified 40b76e7b

[js/web] Forward WebGPU EP buffer cache mode options from JS (#29017)

ssam18 committed 11 days ago

Verified 67c2bc19

[WebGPU] Fix profiling timestamp alignment with ORT profiler (#29021)

daijh committed 11 days ago

Verified cb169366

[CUDA] Fuse MoE router bias into MatMulNBits GEMV (#29170)

tianleiwu committed 11 days ago

Verified ba45260e

[Pipeline] Fix CUDA Python packaging pipeline (#29241)

hariharans29 committed 12 days ago

Verified cf89c494

Fix parsing issue with WebGPU testing skill (#29237)

adrastogi committed 12 days ago

Verified 67aa0aa7

[CUDA] Add sliding-window support to XQA decode (#29177)

tianleiwu committed 12 days ago

Verified 669b8834

Fix CoreML-enabled static build on macOS (#27960)

umireon committed 12 days ago

Verified 9db2150f

Fix incorrect string tensor creation in test (#29219)

yuslepukhin committed 12 days ago

Verified 6e9596e6

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

grybouilli committed 12 days ago

Verified a08e17aa

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

titaiwangms committed 12 days ago

Verified 20ce99fe

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

tairenpiao committed 12 days ago

Verified c69373c4

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

Copilot committed 12 days ago

Verified 701d88ff

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

tianleiwu committed 13 days ago

Verified 1472c16e

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

hariharans29 committed 13 days ago

Verified 8c62bee6

Older