microsoft/onnxruntime

Pull Requests Commits

Fix partitioned capture output types and minimal-build size

chilo-ms committed 3 days ago

e1a48ef6

Fix partitioned CUDA capture ownership and failure handling

chilo-ms committed 4 days ago

a334b87a

Support partitioned CUDA graph capture with the CUDA plugin EP

chilo-ms committed 4 days ago

a9f2334e

Prefer whole-session CUDA graph replay for eligible placement

chilo-ms committed 5 days ago

17ea597b

Prototype CPU offloading with partitioned CUDA graph replay

chilo-ms committed 6 days ago

8a2eb4f4

Add mixed-width weight contract for QMoE (#32697)

jiafatom committed 6 days ago

Verified 090f6ba1

[WebGPU] Refactor subgroup matrix config selection (#32700)

gyagp committed 6 days ago

Verified 30b145bf

Harden shape inference for contrib ops (#32607)

shiyi9801 committed 6 days ago

Verified bca9d15b

[CUDA] Enable non-causal PagedAttention with small cache pages (#32619)

Copilot committed 6 days ago

Verified 51bda5f5

Fix out-of-bounds read when a Loop/Scan body input is also an initializer (#32609)

wangw-1991 committed 8 days ago

Verified 09dfa6ad

Fix heap buffer overflow from reusing a packed sub-byte buffer for a full-byte tensor (#32611)

wangw-1991 committed 8 days ago

Verified 061eecbd

[CPU] Add opt-in MatMulNBits FP32 throughput mode (#31723)

mustjab committed 8 days ago

Verified e69f8415

Add `activation` mode to `com.microsoft::GatedRMSNorm` for SiLU/Swish/Sigmoid across CPU, CUDA, and WebGPU (#32512)

Copilot committed 8 days ago

Verified 0e18025d

Speedup on macos github actions: binary vcpkg cache (#32692)

eserscor committed 9 days ago

Verified e6cd18da

Remove WebGPU EP Python packaging pipeline (#29744)

edgchen1 committed 9 days ago

Verified 1100e116

Collapse A10 pools to a single pool (#32691)

eserscor committed 9 days ago

Verified b55c1c39

Update syntax for release mode in NPM (#32689)

eserscor committed 9 days ago

Verified 46a9c8f5

Fix binary vcpkg caching for github actions (#32682)

eserscor committed 9 days ago

Verified f90e813a

Limit sparse initializer dense allocation (#32684)

apsonawane committed 9 days ago

Verified 62f5193d

Execution plan to offload export on CPU for a MoE (#32261)

xadupre committed 9 days ago

Verified 6b3206e3

Load CUDA external data through pinned buffers (#32437)

xadupre committed 9 days ago

Verified b8df4d66

Combine npm and nuget packing pipelines (#32655)

eserscor committed 9 days ago

Verified ca66cc8f

Add GQA workspace estimation (#32617)

titaiwangms committed 9 days ago

Verified bc8e7ed7

[WebGPU EP] Add native WebGPU GatedDeltaNet kernel (#32510)

Copilot committed 9 days ago

Verified 6014557e

Fix WebGPU subgroup config header dependency (#32672)

xadupre committed 10 days ago

Verified f7d31b39

Remove DML packaging (#32677)

eserscor committed 10 days ago

Verified 20ac0263

Bound windowed GQA workspace allocations (#32602)

titaiwangms committed 10 days ago

Verified 0b487785

add parallelism to prepacking/weight loading (#31691)

xadupre committed 10 days ago

Verified 6f7605c1

Move back to main A10 image (#32676)

eserscor committed 10 days ago

Verified a97735f3

Reject non-control-flow nodes that carry subgraphs during session state finalization (#32641)

miaobin committed 10 days ago

Verified b95cf781

Older