{"repo":"webmachinelearning/webnn","summary":{"retrieved":"2026-05-16T07:49:30Z","triageViolations":21,"urgentViolations":0,"soonViolations":0,"agendaViolations":0,"needsEditsViolations":0,"needTriage":0,"urgent":0,"soon":0,"agenda":0,"needsEdits":0,"other":62},"triage":[{"number":21,"url":"https://github.com/webmachinelearning/webnn/issues/21","title":"(Computer) network-related use cases","author":"anssiko","createdAt":"2019-06-26T17:16:53Z","labels":["use case"],"sloTimeUsed":"P2515DT14H32M37S","whichSlo":"triage","stats":{"numTimelineItems":3,"numComments":1,"numLabels":1},"outOfSlo":true},{"number":24,"url":"https://github.com/webmachinelearning/webnn/issues/24","title":"Backwards compatibility of the initial set of ops","author":"anssiko","createdAt":"2019-08-22T15:05:43Z","labels":["opset"],"sloTimeUsed":"P2458DT16H43M47S","whichSlo":"triage","stats":{"numTimelineItems":3,"numComments":1,"numLabels":1},"outOfSlo":true},{"number":165,"url":"https://github.com/webmachinelearning/webnn/issues/165","title":"Add models to the spec appendix","author":"anssiko","createdAt":"2021-05-06T10:51:37Z","labels":["good first issue","editorial"],"sloTimeUsed":"P1835DT20H57M53S","whichSlo":"triage","stats":{"numTimelineItems":2,"numComments":0,"numLabels":2},"outOfSlo":true},{"number":189,"url":"https://github.com/webmachinelearning/webnn/issues/189","title":"Request for BatchToSpaceND and SpaceToBatchND ops","author":"Honry","createdAt":"2021-07-06T06:17:20Z","labels":["opset","feature request"],"sloTimeUsed":"P1775DT1H32M10S","whichSlo":"triage","stats":{"numTimelineItems":4,"numComments":0,"numLabels":2},"outOfSlo":true},{"number":252,"url":"https://github.com/webmachinelearning/webnn/issues/252","title":"Support for non-IEEE 754 float point types","author":"wacky6","createdAt":"2022-02-10T02:25:04Z","labels":["question","feature request"],"sloTimeUsed":"P1556DT5H24M26S","whichSlo":"triage","stats":{"numTimelineItems":2,"numComments":0,"numLabels":2},"outOfSlo":true},{"number":374,"url":"https://github.com/webmachinelearning/webnn/issues/374","title":"Simplify `MLPool2dOptions` by removing the `outputSizes` option","author":"huningxin","createdAt":"2023-04-04T08:39:42Z","labels":["operator specific"],"sloTimeUsed":"P1137DT23H9M48S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":537,"url":"https://github.com/webmachinelearning/webnn/issues/537","title":"Vision transformers","author":"bhack","createdAt":"2024-01-28T16:08:47Z","labels":["use case"],"sloTimeUsed":"P838DT15H40M43S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":559,"url":"https://github.com/webmachinelearning/webnn/issues/559","title":"Control flow operations: if, while","author":"philloooo","createdAt":"2024-02-06T19:25:35Z","labels":["opset","feature request"],"sloTimeUsed":"P829DT12H23M55S","whichSlo":"triage","stats":{"numTimelineItems":2,"numComments":0,"numLabels":2},"outOfSlo":true},{"number":759,"url":"https://github.com/webmachinelearning/webnn/issues/759","title":"MLOpSupportLimits should be opt-in with base functionality","author":"mwyrzykowski","createdAt":"2024-09-11T23:24:39Z","labels":["policy selection"],"sloTimeUsed":"P611DT8H24M51S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":766,"url":"https://github.com/webmachinelearning/webnn/issues/766","title":"Allow setting label for constant operand","author":"huningxin","createdAt":"2024-10-11T05:15:28Z","labels":["feature request"],"sloTimeUsed":"P582DT2H34M2S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":867,"url":"https://github.com/webmachinelearning/webnn/issues/867","title":"May we update `reversed` options of `cumulativeSum` operation as `reverse`?","author":"BruceDai","createdAt":"2025-06-09T09:57:09Z","labels":["question"],"sloTimeUsed":"P340DT21H52M21S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":876,"url":"https://github.com/webmachinelearning/webnn/issues/876","title":"How to support explicit padding on TFLite backend","author":"fujunwei","createdAt":"2025-08-05T05:23:41Z","labels":["question"],"sloTimeUsed":"P284DT2H25M49S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":877,"url":"https://github.com/webmachinelearning/webnn/issues/877","title":"How to support edge padding mode on TFLite backend","author":"fujunwei","createdAt":"2025-08-05T07:01:28Z","labels":["question"],"sloTimeUsed":"P284DT48M2S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":878,"url":"https://github.com/webmachinelearning/webnn/issues/878","title":"How to support scatterElements on TFLite backend when indices operand is not constant","author":"fujunwei","createdAt":"2025-08-05T07:22:41Z","labels":["question"],"sloTimeUsed":"P284DT26M49S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":896,"url":"https://github.com/webmachinelearning/webnn/issues/896","title":"Add minimum data type set and rank range for input, constant, output and each operator into Spec","author":"BruceDai","createdAt":"2025-10-28T01:59:14Z","labels":[],"sloTimeUsed":"P200DT5H50M16S","whichSlo":"triage","stats":{"numTimelineItems":0,"numComments":0,"numLabels":0},"outOfSlo":true},{"number":899,"url":"https://github.com/webmachinelearning/webnn/issues/899","title":"Idea for Discussion: Incremental Batch Learning and Fine-tuning","author":"AdamSobieski","createdAt":"2025-10-30T18:33:51Z","labels":["question"],"sloTimeUsed":"P197DT13H15M39S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":900,"url":"https://github.com/webmachinelearning/webnn/issues/900","title":"CPU fallback hint","author":"anssiko","createdAt":"2025-10-31T07:38:40Z","labels":["policy selection"],"sloTimeUsed":"P197DT10M50S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":903,"url":"https://github.com/webmachinelearning/webnn/issues/903","title":"Expand the expand operator to support blockwise broadcasting","author":"fdwr","createdAt":"2025-11-10T00:08:00Z","labels":["opset"],"sloTimeUsed":"P187DT7H41M30S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":919,"url":"https://github.com/webmachinelearning/webnn/issues/919","title":"Creating a context should be done using finer-grained in-parallel steps","author":"gterzian","createdAt":"2026-02-17T09:49:20Z","labels":["conventions"],"sloTimeUsed":"P87DT22H10S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":0,"numLabels":1},"outOfSlo":true},{"number":927,"url":"https://github.com/webmachinelearning/webnn/issues/927","title":"Security considerations refresh","author":"anssiko","createdAt":"2026-03-27T08:43:12Z","labels":["security-tracker"],"sloTimeUsed":"P49DT23H6M18S","whichSlo":"triage","stats":{"numTimelineItems":1,"numComments":1,"numLabels":1},"outOfSlo":true},{"number":929,"url":"https://github.com/webmachinelearning/webnn/issues/929","title":"RFE: Add batching support (N-D tensors) to the gemm operator","author":"mtavenrath","createdAt":"2026-04-20T14:11:44Z","labels":[],"sloTimeUsed":"P25DT17H37M46S","whichSlo":"triage","stats":{"numTimelineItems":0,"numComments":0,"numLabels":0},"outOfSlo":true}],"urgent":[],"soon":[],"agenda":[],"needsEdits":[],"other":[{"number":1,"url":"https://github.com/webmachinelearning/webnn/issues/1","title":"Look into pre-canned models","author":"anssiko","createdAt":"2018-11-02T11:59:10Z","labels":["v2"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":6,"url":"https://github.com/webmachinelearning/webnn/issues/6","title":"Custom operations","author":"dsmilkov","createdAt":"2019-01-10T15:32:48Z","labels":["v2","policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":24,"numLabels":2}},{"number":32,"url":"https://github.com/webmachinelearning/webnn/issues/32","title":"Web Neural Network API as-is in WASI?","author":"anssiko","createdAt":"2019-11-06T08:13:57Z","labels":["use case"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":74,"url":"https://github.com/webmachinelearning/webnn/issues/74","title":"[Explainer] explain the workflow from authoring tools to webnn","author":"huningxin","createdAt":"2020-06-23T07:51:32Z","labels":["editorial"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":1,"numLabels":1}},{"number":124,"url":"https://github.com/webmachinelearning/webnn/issues/124","title":"WebNN needs to support models requiring dynamic shape inference","author":"wchao1115","createdAt":"2020-12-08T04:34:09Z","labels":["opset","feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":2,"numLabels":2}},{"number":180,"url":"https://github.com/webmachinelearning/webnn/issues/180","title":"Should dilated pooling be supported","author":"fujunwei","createdAt":"2021-06-08T07:31:04Z","labels":["operator specific","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":8,"numComments":7,"numLabels":2}},{"number":226,"url":"https://github.com/webmachinelearning/webnn/issues/226","title":"Integration with real-time video processing","author":"dontcallmedom","createdAt":"2021-11-10T09:46:39Z","labels":["use case"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":39,"numLabels":1}},{"number":244,"url":"https://github.com/webmachinelearning/webnn/issues/244","title":"A conformance suite with disallowed intra-op examples would be helpful for hardening.","author":"quidity","createdAt":"2021-12-22T10:09:26Z","labels":["security-tracker","testing"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":6,"numLabels":2}},{"number":307,"url":"https://github.com/webmachinelearning/webnn/issues/307","title":"Default axis being 1 of batchNormalization operation is invalid for 1D input tensor","author":"BruceDai","createdAt":"2022-12-08T14:50:21Z","labels":["question","operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":5,"numLabels":2}},{"number":343,"url":"https://github.com/webmachinelearning/webnn/issues/343","title":"Add sample code for WebGPU interop","author":"huningxin","createdAt":"2023-02-15T05:46:32Z","labels":["editorial","webgpu interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":4,"numComments":1,"numLabels":2}},{"number":353,"url":"https://github.com/webmachinelearning/webnn/issues/353","title":"Support `depth_multiplier > 1` for a depthwise conv2d op","author":"Honry","createdAt":"2023-02-27T11:19:21Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":391,"url":"https://github.com/webmachinelearning/webnn/issues/391","title":"Allow 0 size dimensions (dimensions containing a 0 in the list of sizes, not a rank of 0 which is valid)","author":"huningxin","createdAt":"2023-05-23T00:20:11Z","labels":["feature request","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":24,"numLabels":2}},{"number":402,"url":"https://github.com/webmachinelearning/webnn/issues/402","title":"Should we add some model-based tests into WPT","author":"BruceDai","createdAt":"2023-06-20T07:56:56Z","labels":["testing"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":9,"numLabels":1}},{"number":444,"url":"https://github.com/webmachinelearning/webnn/issues/444","title":"Zero dilations needs spec clarification","author":"fdwr","createdAt":"2023-08-02T05:17:25Z","labels":["operator specific","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":2,"numLabels":2}},{"number":462,"url":"https://github.com/webmachinelearning/webnn/issues/462","title":"Improve \"underlying platform\" references in spec","author":"inexorabletash","createdAt":"2023-09-08T00:22:06Z","labels":["conventions"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":470,"url":"https://github.com/webmachinelearning/webnn/issues/470","title":"Simplify `matmul` op","author":"huningxin","createdAt":"2023-10-18T08:51:53Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":10,"numLabels":1}},{"number":480,"url":"https://github.com/webmachinelearning/webnn/issues/480","title":"Proposed use cases from Web LLM project","author":"anssiko","createdAt":"2023-11-16T16:15:17Z","labels":["use case"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":481,"url":"https://github.com/webmachinelearning/webnn/issues/481","title":"Should `scale` and `bias` be required inputs for `batchNormalization` op?","author":"huningxin","createdAt":"2023-11-19T06:33:29Z","labels":["operator specific","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":4,"numLabels":2}},{"number":486,"url":"https://github.com/webmachinelearning/webnn/issues/486","title":"Add \"implementation consideration\" about how out-of-bound indices of Gather/Scatter should be handled","author":"huningxin","createdAt":"2023-12-01T03:03:20Z","labels":["operator specific","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":4,"numComments":5,"numLabels":2}},{"number":487,"url":"https://github.com/webmachinelearning/webnn/issues/487","title":"Should `axes` be required parameter for layerNormalization build method?","author":"huningxin","createdAt":"2023-12-01T04:12:23Z","labels":["operator specific","interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":1,"numLabels":2}},{"number":490,"url":"https://github.com/webmachinelearning/webnn/issues/490","title":"Clarify `instanceNormalization`'s support for 1d, 2d, 3d use cases","author":"huningxin","createdAt":"2023-12-08T06:31:18Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":3,"numLabels":1}},{"number":524,"url":"https://github.com/webmachinelearning/webnn/issues/524","title":"SSMs/Mamba models","author":"bhack","createdAt":"2024-01-21T12:13:56Z","labels":["use case"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":529,"url":"https://github.com/webmachinelearning/webnn/issues/529","title":"Specify WebNN timelines","author":"a-sully","createdAt":"2024-01-25T19:05:53Z","labels":["webgpu interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":573,"url":"https://github.com/webmachinelearning/webnn/issues/573","title":"Core operator set","author":"philloooo","createdAt":"2024-02-16T18:03:19Z","labels":["question","opset"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":4,"numComments":9,"numLabels":2}},{"number":623,"url":"https://github.com/webmachinelearning/webnn/issues/623","title":"WebNN should support NPU and QDQ operations","author":"wchao1115","createdAt":"2024-03-27T00:24:48Z","labels":["opset","feature request","policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":5,"numComments":16,"numLabels":3}},{"number":689,"url":"https://github.com/webmachinelearning/webnn/issues/689","title":"Consider removing `lstm` and `gru` operators","author":"a-sully","createdAt":"2024-05-16T00:58:32Z","labels":["question","operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":15,"numLabels":2}},{"number":691,"url":"https://github.com/webmachinelearning/webnn/issues/691","title":"Divide-by-zero outcome should be standardized","author":"huningxin","createdAt":"2024-05-24T00:31:29Z","labels":["interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":2,"numLabels":1}},{"number":714,"url":"https://github.com/webmachinelearning/webnn/issues/714","title":"Support multiple op sets / builders","author":"zolkis","createdAt":"2024-06-29T06:52:40Z","labels":["question","v2","opset"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":3,"numLabels":3}},{"number":716,"url":"https://github.com/webmachinelearning/webnn/issues/716","title":"MLBuffer destruction timeline","author":"reillyeon","createdAt":"2024-07-02T01:45:45Z","labels":["question"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":14,"numLabels":1}},{"number":748,"url":"https://github.com/webmachinelearning/webnn/issues/748","title":"Is there a reason to support out of order `axes` for layerNormalization?","author":"philloooo","createdAt":"2024-08-01T22:08:44Z","labels":["question","operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":4,"numLabels":2}},{"number":751,"url":"https://github.com/webmachinelearning/webnn/issues/751","title":"Consider separately pass each activation and direction's weights&bias for lstm and gru","author":"philloooo","createdAt":"2024-08-16T18:19:30Z","labels":["question","opset"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":4,"numLabels":2}},{"number":763,"url":"https://github.com/webmachinelearning/webnn/issues/763","title":"Request standards positions from Mozilla and WebKit","author":"reillyeon","createdAt":"2024-09-19T21:40:40Z","labels":["process"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":12,"numLabels":1}},{"number":780,"url":"https://github.com/webmachinelearning/webnn/issues/780","title":"Idea for Discussion: Device Memory Management Primitives","author":"wacky6","createdAt":"2024-11-07T06:52:53Z","labels":["feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":794,"url":"https://github.com/webmachinelearning/webnn/issues/794","title":"Disallow operations on scalar tensors that are no-ops","author":"reillyeon","createdAt":"2024-11-23T01:46:25Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":9,"numLabels":1}},{"number":807,"url":"https://github.com/webmachinelearning/webnn/issues/807","title":"Caching mechanism for MLGraph","author":"anssiko","createdAt":"2025-01-22T11:34:01Z","labels":["question","feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":37,"numLabels":2}},{"number":821,"url":"https://github.com/webmachinelearning/webnn/issues/821","title":"Operator naming 2D vs 2d","author":"fdwr","createdAt":"2025-02-21T04:50:59Z","labels":["conventions"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":822,"url":"https://github.com/webmachinelearning/webnn/issues/822","title":"(de)quantization behaviors on CoreML","author":"philloooo","createdAt":"2025-02-21T23:45:06Z","labels":["interop"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":12,"numLabels":1}},{"number":853,"url":"https://github.com/webmachinelearning/webnn/issues/853","title":"The minimum data type set","author":"huningxin","createdAt":"2025-05-28T09:26:53Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":856,"url":"https://github.com/webmachinelearning/webnn/issues/856","title":"Consider drop the support for int32/uint32 of zeropoint for quantizeLinear","author":"lisa0314","createdAt":"2025-06-03T05:35:51Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":8,"numLabels":1}},{"number":861,"url":"https://github.com/webmachinelearning/webnn/issues/861","title":"Evaluate sustainability impact","author":"anssiko","createdAt":"2025-06-06T08:55:45Z","labels":["tag-tracker"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":5,"numComments":7,"numLabels":1}},{"number":870,"url":"https://github.com/webmachinelearning/webnn/issues/870","title":"Improve accessibility of examples","author":"anssiko","createdAt":"2025-06-24T09:07:25Z","labels":["editorial"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":4,"numLabels":1}},{"number":874,"url":"https://github.com/webmachinelearning/webnn/issues/874","title":"Normalize the behavior when NaN is used for minValue or maxValue of clamp operator","author":"wangw-1991","createdAt":"2025-08-04T01:24:48Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":1,"numLabels":1}},{"number":879,"url":"https://github.com/webmachinelearning/webnn/issues/879","title":"Propose to remove WPT conformance tests with negative scale of dequantizeLinear and quantizeLinear operator","author":"BruceDai","createdAt":"2025-08-05T08:38:06Z","labels":["question"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":2,"numLabels":1}},{"number":883,"url":"https://github.com/webmachinelearning/webnn/issues/883","title":"Support flexible input sizes","author":"huningxin","createdAt":"2025-08-27T08:51:32Z","labels":["feature request","operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":8,"numComments":19,"numLabels":2}},{"number":897,"url":"https://github.com/webmachinelearning/webnn/issues/897","title":"Define \"underlying execution device\" concept","author":"anssiko","createdAt":"2025-10-28T10:45:52Z","labels":["policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":1,"numLabels":1}},{"number":898,"url":"https://github.com/webmachinelearning/webnn/issues/898","title":"Support for workloads with response time requirements (realtime)","author":"handellm","createdAt":"2025-10-29T10:25:46Z","labels":[],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":2,"numComments":4,"numLabels":0}},{"number":901,"url":"https://github.com/webmachinelearning/webnn/issues/901","title":"Proposal: API to Separate Graph Building from Weight Loading to Reduce Peak Memory Usage","author":"mtavenrath","createdAt":"2025-11-07T07:51:43Z","labels":["feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":7,"numLabels":1}},{"number":902,"url":"https://github.com/webmachinelearning/webnn/issues/902","title":"Device selection criteria for usecase-driven scenarios","author":"fdwr","createdAt":"2025-11-09T23:55:21Z","labels":["policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":11,"numLabels":1}},{"number":904,"url":"https://github.com/webmachinelearning/webnn/issues/904","title":"2D, or not 2D, that is the question","author":"fdwr","createdAt":"2025-11-10T00:52:27Z","labels":["opset"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":3,"numLabels":1}},{"number":907,"url":"https://github.com/webmachinelearning/webnn/issues/907","title":"Composite operators / subgraphs","author":"fdwr","createdAt":"2025-11-27T01:01:19Z","labels":["opset"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":911,"url":"https://github.com/webmachinelearning/webnn/issues/911","title":"accelerated should be prior to powerPreference for device selection","author":"mingmingtasd","createdAt":"2025-12-18T07:36:00Z","labels":["policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":32,"numLabels":1}},{"number":914,"url":"https://github.com/webmachinelearning/webnn/issues/914","title":"Floating point accuracy for sin and cos with range constraint","author":"philloooo","createdAt":"2026-01-05T22:58:32Z","labels":["operator specific"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":6,"numLabels":1}},{"number":924,"url":"https://github.com/webmachinelearning/webnn/issues/924","title":"Consider specifying tensor element count limit","author":"philloooo","createdAt":"2026-03-19T17:44:43Z","labels":["feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":4,"numLabels":1}},{"number":925,"url":"https://github.com/webmachinelearning/webnn/issues/925","title":"Mandate `output_channels % groups == 0` validation for conv2d","author":"lynne0326","createdAt":"2026-03-20T17:23:30Z","labels":["security-tracker"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":1,"numLabels":1}},{"number":928,"url":"https://github.com/webmachinelearning/webnn/issues/928","title":"Consider specify lower limit for conv2d/pool2d kernel sizes, dilations, strides","author":"philloooo","createdAt":"2026-04-02T23:00:29Z","labels":["security-tracker"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":5,"numComments":13,"numLabels":1}},{"number":930,"url":"https://github.com/webmachinelearning/webnn/issues/930","title":"RFE:  Add support for more floating point low-precision ML data types (`bfloat16`, `fp8`, `nvfp4`)","author":"mtavenrath","createdAt":"2026-04-21T14:02:33Z","labels":["opset","feature request"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":4,"numComments":8,"numLabels":2}},{"number":931,"url":"https://github.com/webmachinelearning/webnn/issues/931","title":"Define an upper bound on number of concat inputs and split outputs","author":"philloooo","createdAt":"2026-04-24T17:59:41Z","labels":["security-tracker"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":3,"numComments":5,"numLabels":1}},{"number":934,"url":"https://github.com/webmachinelearning/webnn/issues/934","title":"Effective MLComputePolicy exposure","author":"anssiko","createdAt":"2026-05-07T16:12:29Z","labels":["policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":1,"numComments":2,"numLabels":1}},{"number":910,"url":"https://github.com/webmachinelearning/webnn/pull/910","title":"add minimum data types and rank range for operations","author":"BruceDai","createdAt":"2025-12-10T09:43:21Z","pull_request":{"draft":false},"labels":[],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":77,"numComments":77,"numLabels":0}},{"number":923,"url":"https://github.com/webmachinelearning/webnn/pull/923","title":"Refactor device selection: Rename to computePolicy, remove accelerated, and add fallback","author":"mingmingtasd","createdAt":"2026-03-18T06:32:28Z","pull_request":{"draft":false},"labels":["policy selection"],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":35,"numComments":35,"numLabels":1}},{"number":932,"url":"https://github.com/webmachinelearning/webnn/pull/932","title":"Validate output channels are a multiple of groups in conv2d","author":"lynne0326","createdAt":"2026-05-04T20:21:57Z","pull_request":{"draft":false},"labels":[],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":5,"numComments":5,"numLabels":0}},{"number":933,"url":"https://github.com/webmachinelearning/webnn/pull/933","title":"validate concat inputs and split splits","author":"philloooo","createdAt":"2026-05-06T17:50:19Z","pull_request":{"draft":false},"labels":[],"sloTimeUsed":"PT0S","whichSlo":"none","stats":{"numTimelineItems":8,"numComments":8,"numLabels":0}}]}