microsoft/onnxruntime

Pull Requests Commits

more fusion in llm - propose GemmQuickGelu

guschmue committed 283 days ago

e90c7264

Enable multithreading on FP16 to FP32 cast operator (#23619)

Erick Muñoz committed 283 days ago

Verified 7fc7d5ec

[WebNN/WebGPU JS] Fix shared Module methods overriding each other (#23998)

egalli committed 283 days ago

Verified eceae8b2

Add macOS ARM64 pipeline for webgpu (#24060)

fs-eire committed 283 days ago

Verified d98046b3

[webgpu] Limit that K must be divisible by 128 to apply dp4a matmul (#24078)

qjia7 committed 283 days ago

Verified a85977dd

Improve typing for OrtValue and other public Python interfaces (#24086)

justinchuby committed 284 days ago

Verified 12fea572

Fix CUDA EP Abs and Sign bfloat16 support (#23914)

skottmckay committed 284 days ago

Verified c6a26754

Upgrade current MacOS-13 to 14 (#23293)

jchen351 committed 284 days ago

Verified e495750a

[WebGPU EP] fix for reduce min/max error on MacOS CI (#24077)

prathikr committed 284 days ago

Verified b3aa5a3c

add bool support to EPContext schema to unblock some models (#24065)

HectorSVC committed 284 days ago

Verified a46d2127

[WebGPU EP] Implement Remaining Reduction Ops (#24045)

prathikr committed 285 days ago

Verified 154e3b7d

[WebGPU] Reduce staging buffers for uploading intializers (#23968)

jchen10 committed 285 days ago

Verified f5812d0e

Extend CMAKE_CUDA_FLAGS with all Blackwell compute capacity (#23928)

yf711 committed 285 days ago

Verified 2bc73ca8

Windows: Show more useful DLL load errors to say exactly what DLL is missing (#24053)

RyanUnderhill committed 286 days ago

Verified 5ef0d211

Whisper Redesigned Solution (#23549)

kunal-vaishnavi committed 287 days ago

Verified 7942fa7a

[Native WebGPU EP] Add packedQKV and do_rotary attribute support to GroupQueryAttention operator (#23386)

satyajandhyala committed 288 days ago

Verified ae501eeb

[ARM CPU] Fix fp16 const initialization on no-fp16 platform (#23978)

fajin-corp committed 288 days ago

Verified f22ee08f

Convert Windows GPU pipelines and Windows OpenVino pipeline to Github Actions (#24029)

Changming Sun committed 288 days ago

Verified b8966665

[WebNN] Better int64 integration (#23831)

Honry committed 288 days ago

Verified 73d9826a

Add support for custom position ids and attention bias to GQA CPU operator (#23944)

derdeljan-msft committed 288 days ago

Verified 5a694bcb

avoid copy unnecessary files for nodejs pkg (#23992)

fs-eire committed 288 days ago

Verified 41c239df

[webgpu] fix the wrong dispatch size in flash_attention (#24020)

xhcao committed 288 days ago

Verified cc5840be

Move call to MLAS_CPUIDINFO::GetCPUIDInfo() out of MlasSQNBitGemmDispatchNeon initialization. (#24018)

edgchen1 committed 288 days ago

Verified c9c8b48d

[webgpu/wasm] allow runtime switch between WebGPUEP and JSEP (#24032)

fs-eire committed 288 days ago

Verified 7c05e7f5

Make Cuda packaging pipeline 1ES compliant (#23806)

jchen351 committed 288 days ago

Verified 9f214561

Set CMAKE_POLICY_DEFAULT_CMP0069 to NEW to ensure that IPO flags are added for dependencies. (#24034)

edgchen1 committed 288 days ago

Verified 9a296a0a

[TensorRT EP] Stop enforcing oss parser during Windows debug build (#24036)

yf711 committed 288 days ago

Verified cb3f631f

Support tvOS build (#24000)

hans00 committed 288 days ago

Verified 1fc6d8ca

Increase timeout for ARM64-Xcode16-targeting-iphonesimulator (#24030)

edgchen1 committed 289 days ago

Verified 3f71d637

[TensorRT EP] Call cudaSetDevice at compute function for handling multithreading scenario (#24010)

chilo-ms committed 289 days ago

Verified 6bb6d791

Older