Pull requests · openvinotoolkit/openvino

[GGUF] Store Q3_K weights as OpenVINO u3

#38933mvafin opened

Oct 16, 2026

Contributor

[NPUW] Suppport 4d float attention_mask

#38932OrestChura opened

Oct 16, 2026

Contributor

[CPU] Make f32 Mod and FloorMod exact

#38931usmanmateen opened

Oct 16, 2026

Contributor

[benchmark_app] Include ordinary NC inputs in automatic file mapping

#38928zorox06 opened

Oct 16, 2026

[benchmark_app] Recognize an existing batch axis without reshaping

#38927zorox06 opened

Oct 16, 2026

[benchmark_app] Preserve scalar rank for bf16 and packed random inputs

#38926zorox06 opened

Oct 16, 2026

[benchmark_app] Round iteration counts to complete shape groups

#38925zorox06 opened

Oct 16, 2026

[benchmark_app] Quote delimiters and newlines in CSV report fields

#38924zorox06 opened

Oct 16, 2026

[benchmark_app] Keep every averaged profiling counter without mutating requests

#38923zorox06 opened

Oct 16, 2026

[benchmark_app] Fill image info with each width and height pair

#38922zorox06 opened

Oct 16, 2026

[benchmark_app] Preserve the channel axis in dynamic grayscale inputs

#38921zorox06 opened

Oct 16, 2026

[benchmark_app] Fill image tensors without a batch dimension

#38920zorox06 opened

Oct 16, 2026

[benchmark_app] Load NumPy inputs along the declared batch axis

#38919zorox06 opened

Oct 16, 2026

[benchmark_app] Preserve binary tensor values when filling batches

#38918zorox06 opened

Oct 16, 2026

[PT FE] Reduce memory consumption for MoE model conversion

#38917mvafin opened

Oct 16, 2026

Contributor

Model v2 fix

#38916kkoryun opened

Oct 16, 2026

Contributor

[DO NOT MERGE] Update compiler

#38915ArtemySkrebkov opened

Oct 16, 2026

Contributor

1

[NPUW] Validate imported prefill/generate limits against port shapes

#38914darktears opened

Oct 16, 2026

Contributor

[ONNX FE] Unsupported operation DeformConv

#38909bumbosiepsak opened

Oct 16, 2026

Contributor

2

Ubuntu 26.04 tests in GHA workflows

#38908ksankiew opened

Oct 16, 2026

Contributor

[NPU] Check table ext versions only after the plugin negotiates with the driver

#38907pereanub opened

Oct 16, 2026

Contributor

Fix MOECompressed losing MOE base config on deserialize/clone

#38906intelgaoxiong opened

Oct 16, 2026

Contributor

3

[GPU][Tests] Use new shape infer in fused_node.port_from_peer_node

#38905wilson-seok opened

Oct 16, 2026

Contributor

[GPU] Fix key over-read in sdpa_opt partial key block

#38904wilson-seok opened

Oct 16, 2026

Contributor

[GPU] Fix MoE router over-read and mlp_reduce local size larger than global size

#38903wilson-seok opened

Oct 16, 2026

Contributor

Navigation Menu

Repository navigation

All pull requests

[GGUF] Store Q3_K weights as OpenVINO u3

[NPUW] Suppport 4d float attention_mask

[CPU] Make f32 Mod and FloorMod exact

[benchmark_app] Include ordinary NC inputs in automatic file mapping

[benchmark_app] Recognize an existing batch axis without reshaping

[benchmark_app] Preserve scalar rank for bf16 and packed random inputs

[benchmark_app] Round iteration counts to complete shape groups

[benchmark_app] Quote delimiters and newlines in CSV report fields

[benchmark_app] Keep every averaged profiling counter without mutating requests

[benchmark_app] Fill image info with each width and height pair

[benchmark_app] Preserve the channel axis in dynamic grayscale inputs

[benchmark_app] Fill image tensors without a batch dimension

[benchmark_app] Load NumPy inputs along the declared batch axis

[benchmark_app] Preserve binary tensor values when filling batches

[PT FE] Reduce memory consumption for MoE model conversion

Model v2 fix

[DO NOT MERGE] Update compiler

[NPUW] Validate imported prefill/generate limits against port shapes

[ONNX FE] Unsupported operation DeformConv

Ubuntu 26.04 tests in GHA workflows

[NPU] Check table ext versions only after the plugin negotiates with the driver

Fix MOECompressed losing MOE base config on deserialize/clone

[GPU][Tests] Use new shape infer in fused_node.port_from_peer_node

[GPU] Fix key over-read in sdpa_opt partial key block

[GPU] Fix MoE router over-read and mlp_reduce local size larger than global size