microsoft/onnxruntime

Pull Requests Commits

Update softmax test

derdeljan-msft committed 1 year ago

26c0a9f1

Store masked input in the output buffer

derdeljan-msft committed 1 year ago

23223fa7

Upcast bias to fp32 if needed

derdeljan-msft committed 1 year ago

c3396ab8

derdeljan-msft committed 1 year ago

3e56b0d9

Softmax bias add + reduce max fp32 kernel

derdeljan-msft committed 1 year ago

8ff5a947

Whisper Redesigned Solution (#23549)

kunal-vaishnavi committed 1 year ago

Verified 7942fa7a

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

satyajandhyala committed 1 year ago

Verified ae501eeb

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

fajin-corp committed 1 year ago

Verified f22ee08f

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

Changming Sun committed 1 year ago

Verified b8966665

[WebNN] Better int64 integration (#23831)

Honry committed 1 year ago

Verified 73d9826a

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

derdeljan-msft committed 1 year ago

Verified 5a694bcb

avoid copy unnecessary files for nodejs pkg (#23992)

fs-eire committed 1 year ago

Verified 41c239df

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

xhcao committed 1 year ago

Verified cc5840be

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

edgchen1 committed 1 year ago

Verified c9c8b48d

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

fs-eire committed 1 year ago

Verified 7c05e7f5

Make Cuda packaging pipeline 1ES compliant (#23806)

jchen351 committed 1 year ago

Verified 9f214561

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

edgchen1 committed 1 year ago

Verified 9a296a0a

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

yf711 committed 1 year ago

Verified cb3f631f

Support tvOS build (#24000)

hans00 committed 1 year ago

Verified 1fc6d8ca

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

edgchen1 committed 1 year ago

Verified 3f71d637

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

chilo-ms committed 1 year ago

Verified 6bb6d791

Remove unused parameter in csharp InferenceTest (#24031)

tianleiwu committed 1 year ago

Verified 5bd31636

Fix attention bias broadcast (#24017)

tianleiwu committed 1 year ago

Verified 99b78a94

fix test for RotaryEmbedding (#24022)

fs-eire committed 1 year ago

Verified 219c919c

[webgpu] allow overloads to Program::AddIndices (#24021)

fs-eire committed 1 year ago

Verified 401f24a8

[webgpu] allow to specify UseIndicesTypeAlias for Indices (#24019)

fs-eire committed 1 year ago

Verified 1362e7ca

Correct generated cmake syntax (#24016)

hans00 committed 1 year ago

Verified 11216a4e

Qnn weight sharing improvement (#23945)

HectorSVC committed 1 year ago

Verified 7ae606f7

Add DNNL github workflow (#24011)

Changming Sun committed 1 year ago

Verified 57ddd026

Update onnxruntime_c_api.h to work with MinGW (#24006)

hsilm committed 1 year ago

Verified 5e057292

Older