Commits microsoft/onnxruntime

KV cache quantization for ORT WebGPU (#28059)

sushraja-msft committed 12 days ago

Verified d9fcce02

Fix FP8 (FLOAT8E4M3FN) quantization scale using wrong reference distribution (#29350)

Osamaali313 committed 14 days ago

Verified 6c522b4d

[CUDA] Remove unused code in moe_kernels.cu‎ (#29295)

tianleiwu committed 14 days ago

Verified c3a5222d

Normalize bool tensor raw_data to {0, 1} on unpack (#29238)

jiafatom committed 14 days ago

Verified c65d9b64

Crypto support : App supply I/O callbacks to EP + callback and fallback helpers (#28624)

GopalakrishnanN committed 15 days ago

Verified 49efb328

Fix CUDA/cuDNN DLL preload paths for CUDA 13 consolidated wheel layout (#29202)

tianleiwu committed 15 days ago

Verified 37cccc89

Fix integer overflow in RKNPU implicit bias allocation (#29249)

GopalakrishnanN committed 15 days ago

Verified d857f77a

[CPU] Enable pre-packed weights sharing for MatMulNBits (#29163)

derdeljan-msft committed 15 days ago

Verified a5d2663b

Fix bounds in WhisperDecoderSubgraph::CreateInitialFeeds initial feeds (#29239)

jiafatom committed 15 days ago

Verified 37a3b51c

Validate SparseAttention CSR indices and key lengths element values (#29242)

yuslepukhin committed 15 days ago

Verified 57865882

[CUDA Plugin EP] Expose kernel sync stream for scratch allocation (#29244)

tianleiwu committed 15 days ago

Verified 126ea8d3

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

sanaa-hamel-microsoft committed 15 days ago

Verified 5f49a372

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

sayanshaw24 committed 15 days ago

Verified 0271287d

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

tianleiwu committed 15 days ago

Verified a203dfaf

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

qjia7 committed 15 days ago

Verified 92b4c663

Fix CPU Attention causal mask alignment (#29050)

FuZoe committed 16 days ago

Verified d4b01e43

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

tianleiwu committed 17 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 17 days ago

Verified 4ed5a4ad

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

tianleiwu committed 17 days ago

Verified 996cea1e

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

jiafatom committed 17 days ago

Verified eade9ea9

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

adrastogi committed 17 days ago

Verified b77b3123

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

tianleiwu committed 17 days ago

Verified 40b76e7b

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

ssam18 committed 17 days ago

Verified 67c2bc19

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

daijh committed 17 days ago

Verified cb169366

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

tianleiwu committed 17 days ago

Verified ba45260e

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

hariharans29 committed 18 days ago

Verified cf89c494

Fix parsing issue with WebGPU testing skill (#29237)

adrastogi committed 18 days ago

Verified 67aa0aa7

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

tianleiwu committed 18 days ago

Verified 669b8834

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

umireon committed 18 days ago

Verified 9db2150f

Fix incorrect string tensor creation in test (#29219)

yuslepukhin committed 18 days ago

Verified 6e9596e6