microsoft/onnxruntime

Pull Requests Commits

rotary fully implemented

aciddelgado committed 2 years ago

6e6ad2cb

Merge branch 'yufeng/gqa_opt' into aciddelgado/gqa_rotary

aciddelgado committed 2 years ago

791bbc3d

Merge branch 'main' into aciddelgado/gqa_rotary

aciddelgado committed 2 years ago

ae34d0cf

aciddelgado committed 2 years ago

082f347d

Update gradient ops tests (#18783)

askhade committed 2 years ago

Verified 487abcd2

Fix a build warning in SparseTensor code for 32-bit build configs (#18766)

Changming Sun committed 2 years ago

Verified 17eaf9b0

Move NuGet nightly package publishing job to a separated pipeline (#18801)

Changming Sun committed 2 years ago

Verified 44054e75

[js/webgpu] Provide a naive vectorized matmul algorithm (#18758)

qjia7 committed 2 years ago

Verified b30e721d

Add DynamicQuantizeLinear as supported OP (#18798)

TedThemistokleous committed 2 years ago

Verified 1ad6eb13

Disable test_bert_result_with_layerwise_recompute (#18800)

pengwa committed 2 years ago

Verified dbe886ab

[ROCm] Refactor to hide ck layout (Row/Col) from ORT interface (#18777)

cloudhan committed 2 years ago

Verified 3940ef20

[JS/Web] Added uniforms to Reduce, Resize and Split Ops. (#18727)

satyajandhyala committed 2 years ago

Verified 0ca84549

[QNN EP Quantization] Add fusion preprocessing to QNN quantization (#18719)

adrianlizarraga committed 2 years ago

Verified 81796a30

[PowerPC] Type casting the output operand of vec_xst. (#18057)

BODAPATIMAHESH committed 2 years ago

Verified 65300610

[JS/WebGPU] Added uniforms to Tile and Where Ops (#18768)

satyajandhyala committed 2 years ago

Verified d673e39a

[js/webgpu] Fix shader compilation errors in cumsum (#18779)

qjia7 committed 2 years ago

Verified b4be9e1b

Log out ORT session options (#16259)

ivberg committed 2 years ago

Verified a85ef652

[js/web/training] lazyResetGrad implementation (#18711)

carzh committed 2 years ago

Verified eb030329

Allow layer-wise recompute (#18566)

pengwa committed 2 years ago

Verified ccf3b205

Fix buffer overrun in 4b dequant cuda (#18780)

chenfucn committed 2 years ago

Verified 68c832d5

Adding a new pipeline for publishing to Python Cuda 12 packages. (#18712)

jchen351 committed 2 years ago

Verified ce1fed6d

Adding a new pipeline for pubilshing cuda 12 nuget packages (#18713)

jchen351 committed 2 years ago

Verified bfa5eb45

Update transformers package to fix the security issue (#18730)

askhade committed 2 years ago

Verified 16df8377

Fix GQA shape inference (#18723)

PatriceVignola committed 2 years ago

Verified 8d641229

[ROCm] Add GemmFloat8 (#18488)

cloudhan committed 2 years ago

Verified de32baee

Extend API page on the python documentation (#18762)

xadupre committed 2 years ago

Verified d41dd772

Add cuda visible devices for Mistral benchmark (#18764)

ajindal1 committed 2 years ago

Verified 2f93d97f

Build fixes for Windows ARM32 desktop build (#18752)

Changming Sun committed 2 years ago

Verified c7799d70

Fix gemm_float8 build failure on CUDA 11.3-11.7 (#18760)

pengwa committed 2 years ago

Verified 44b58437

[WebNN EP] Don't covert all inputs except the 0th input for Resize (#18687)

Honry committed 2 years ago

Verified e8f33b54

Older