Web Machine Learning Working Group - Activity in 2025
No matches.
TR publications
-
2 documents published.
-
Web Neural Network API published twice as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
6 participants joined.
-
liang zeng (ByteDance) joined.
-
JuGuang Liu (ByteDance) joined.
-
Victor Huang (Microsoft Corporation) joined.
-
Doug Schepers (W3C Invited Experts) joined.
-
Vasilii Trofimchuk (Block, Inc.) joined.
-
Dexter Yang (ByteDance) joined.
-
-
1 participant left.
-
Brent Zundel (Tradeverifyd) left.
-
-
1 organization joined.
-
Block, Inc. joined.
-
GitHub
webmachinelearning/webnn
-
1 issue opened.
-
accelerated should be prior to powerPreference for device selection webmachinelearning/webnn#911, by @mingmingtasd.
-
-
8 issues received 19 new comments.
-
Device selection criteria for usecase-driven scenarios webmachinelearning/webnn#902 (5 comments by @anssiko, @handellm, @mwyrzykowski, @zolkis)
-
Proposal: API to Separate Graph Building from Weight Loading to Reduce Peak Memory Usage webmachinelearning/webnn#901 (4 comments by @fdwr, @kainino0x, @mtavenrath, @reillyeon)
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836 (2 comments by @philloooo, @zolkis)
-
Request standards positions from Mozilla and WebKit webmachinelearning/webnn#763 (2 comments by @reillyeon, @RReverser)
-
webmachinelearning/webnn#106 (1 comment by @fdwr)
-
-
2 pull requests opened.
-
add minimum data types and rank range for operations webmachinelearning/webnn#910, by @BruceDai.
-
Clarify division operator rounding webmachinelearning/webnn#909, by @fdwr.
-
-
2 pull requests merged.
-
Clarify division operator rounding webmachinelearning/webnn#909.
-
Update WebNN architecture diagram webmachinelearning/webnn#908.
-
-
5 pull requests received 82 new comments.
-
add minimum data types and rank range for operations webmachinelearning/webnn#910 (73 comments by @BruceDai, @fdwr, @huningxin, @reillyeon)
-
Clarify division operator rounding webmachinelearning/webnn#909 (5 comments by @anssiko, @dontcallmedom, @fdwr)
-
Restrict scale's values of dequantizeLinear and quantizeLinear to be positive and nonzero webmachinelearning/webnn#906 (2 comments by @fdwr, @huningxin)
-
Past meetings
Participants
-
10 participants joined.
-
Sharanya Chandrasekaran (PayPal) joined.
-
Noormina Abuthahir (PayPal) joined.
-
Pavan Yanamadala (PayPal) joined.
-
Siddharth Mangesh Gothoskar (PayPal) joined.
-
JaEun Jemma Ku (University of Illinois) joined.
-
Chris Needham (British Broadcasting Corporation) joined.
-
Zoltan Kis (W3C Invited Experts) joined.
-
Ben Greenstein (Google LLC) joined.
-
Lynne Jiang (Google LLC) joined.
-
Simon Wijckmans (Client side development Inc) joined.
-
-
3 organizations joined.
-
PayPal joined.
-
University of Illinois joined.
-
Client side development Inc joined.
-
GitHub
webmachinelearning/webnn
-
5 issues opened.
-
Composite operators / subgraphs webmachinelearning/webnn#907, by @fdwr.
-
2D, or not 2D, that is the question webmachinelearning/webnn#904, by @fdwr.
-
Expand the expand operator to support blockwise broadcasting webmachinelearning/webnn#903, by @fdwr.
-
Device selection criteria for usecase-driven scenarios webmachinelearning/webnn#902, by @fdwr.
-
Proposal: API to Separate Graph Building from Weight Loading to Reduce Peak Memory Usage webmachinelearning/webnn#901, by @mtavenrath.
-
-
2 issues closed.
-
Future-proofing privacy considerations of New API feature to identify feature support per operator webmachinelearning/webnn#875.
-
Revise privacy considerations webmachinelearning/webnn#886.
-
-
9 issues received 24 new comments.
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836 (7 comments by @anssiko, @fdwr, @handellm, @mwyrzykowski, @philloooo, @zolkis)
-
Device selection criteria for usecase-driven scenarios webmachinelearning/webnn#902 (5 comments by @fdwr, @handellm, @mwyrzykowski, @zolkis)
-
Support for workloads with response time requirements (realtime) webmachinelearning/webnn#898 (3 comments by @anssiko, @handellm, @mwyrzykowski)
-
Proposal: API to Separate Graph Building from Weight Loading to Reduce Peak Memory Usage webmachinelearning/webnn#901 (3 comments by @anssiko, @mtavenrath)
-
Future-proofing privacy considerations of New API feature to identify feature support per operator webmachinelearning/webnn#875 (1 comment by @anssiko)
-
Propose to remove WPT conformance tests with negative scale of dequantizeLinear and quantizeLinear operator webmachinelearning/webnn#879 (1 comment by @reillyeon)
-
Request standards positions from Mozilla and WebKit webmachinelearning/webnn#763 (1 comment by @anssiko)
-
-
2 pull requests opened.
-
Update WebNN architecture diagram webmachinelearning/webnn#908, by @anssiko.
-
Restrict scale's values of dequantizeLinear and quantizeLinear to be positive and nonzero webmachinelearning/webnn#906, by @BruceDai.
-
-
1 pull request merged.
-
Revise privacy considerations webmachinelearning/webnn#890.
-
-
4 pull requests received 20 new comments.
-
Restrict scale's values of dequantizeLinear and quantizeLinear to be positive and nonzero webmachinelearning/webnn#906 (10 comments by @BruceDai, @fdwr, @philloooo, @reillyeon)
-
Add a simple accelerator selection mechanism. webmachinelearning/webnn#895 (3 comments by @anssiko, @fdwr, @mtavenrath)
-
webmachinelearning/webnn#905 (1 comment by @anssiko)
-
TR publications
-
1 document published.
-
Web Neural Network API published as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
5 participants joined.
-
Haoli Chen (ByteDance) joined.
-
Sword Li (Cybozu) joined.
-
Cybozu W3C (Cybozu) joined.
-
Stephen Erickson (The Washington Post) joined.
-
Davis Shaver (The Washington Post) joined.
-
-
1 participant left.
-
Zoltan Kis (W3C Invited Experts) left.
-
-
2 organizations joined.
-
Cybozu joined.
-
The Washington Post joined.
-
GitHub
webmachinelearning/webnn
-
4 issues opened.
-
CPU fallback hint webmachinelearning/webnn#900, by @anssiko.
-
Support for workloads with response time requirements (realtime) webmachinelearning/webnn#898, by @handellm.
-
Define "underlying execution device" concept webmachinelearning/webnn#897, by @anssiko.
-
Add minimum data type set and rank range for input, constant, output and each operator into Spec webmachinelearning/webnn#896, by @BruceDai.
-
-
1 issue closed.
-
Query supported devices before graph compilation webmachinelearning/webnn#815.
-
-
8 issues received 13 new comments.
-
Evaluate sustainability impact webmachinelearning/webnn#861 (4 comments by @AlexDawsonUK, @anssiko, @mgifford)
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836 (2 comments by @handellm, @mwyrzykowski)
-
Support for workloads with response time requirements (realtime) webmachinelearning/webnn#898 (1 comment by @handellm)
-
The decomposition of `lstm` has issue for batch size 1 input webmachinelearning/webnn#889 (1 comment by @fdwr)
-
Drop support of 8-bit integers input for CumulativeSum webmachinelearning/webnn#892 (1 comment by @fdwr)
-
-
1 pull request opened.
-
Add a simple accelerator selection mechanism. webmachinelearning/webnn#895, by @zolkis.
-
-
2 pull requests merged.
-
Update explainer with new proposal for simple accelerator mapping webmachinelearning/webnn#884.
-
Add a simple accelerator selection mechanism. webmachinelearning/webnn#895.
-
-
4 pull requests received 37 new comments.
-
Add a simple accelerator selection mechanism. webmachinelearning/webnn#895 (28 comments by @anssiko, @handellm, @huningxin, @reillyeon, @zolkis)
-
Revise privacy considerations webmachinelearning/webnn#890 (5 comments by @anssiko, @reillyeon, @sandandsnow)
-
Update explainer with new proposal for simple accelerator mapping webmachinelearning/webnn#884 (2 comments by @anssiko, @zolkis)
-
TR publications
-
5 documents published.
-
Web Neural Network API published 5 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
6 participants joined.
-
Umar Iqbal (W3C Invited Experts) joined.
-
Mark Foltz (Google LLC) joined.
-
Jason Mayes (Google LLC) joined.
-
Ilya Grigorik (Shopify) joined.
-
Fabio Bernardon (NVIDIA) joined.
-
sandeep kumar (NVIDIA) joined.
-
-
4 participants left.
-
Domenic Denicola (Google LLC) left.
-
Joshua Bell (Google LLC) left.
-
Ilya Rezvov (Google LLC) left.
-
Michael McCool (Intel Corporation) left.
-
-
1 organization joined.
-
Shopify joined.
-
GitHub
webmachinelearning/webnn
-
6 issues opened.
-
Pad operator algorithm steps should not throw error for scalar input webmachinelearning/webnn#893, by @huningxin.
-
Drop support of 8-bit integers input for CumulativeSum webmachinelearning/webnn#892, by @huningxin.
-
The decomposition of `lstm` has issue for batch size 1 input webmachinelearning/webnn#889, by @fujunwei.
-
On-device training? webmachinelearning/webnn#887, by @xbwwj.
-
Revise privacy considerations webmachinelearning/webnn#886, by @anssiko.
-
Support dynamic tensor resizing for slice and resample2d webmachinelearning/webnn#885, by @Honry.
-
-
6 issues closed.
-
Pad operator algorithm steps should not throw error for scalar input webmachinelearning/webnn#893.
-
Support uint8/int8 input for resample2d webmachinelearning/webnn#872.
-
Support dynamic tensor resizing for slice and resample2d webmachinelearning/webnn#885.
-
opSupportLimits: Level of detail for output tensor(s)? webmachinelearning/webnn#835.
-
Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#324.
-
On-device training? webmachinelearning/webnn#887.
-
-
1 issue reopened.
-
Future-proofing privacy considerations of New API feature to identify feature support per operator webmachinelearning/webnn#875, by @anssiko.
-
-
9 issues received 21 new comments.
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (7 comments by @handellm, @mwyrzykowski, @zolkis)
-
Support dynamic tensor resizing for slice and resample2d webmachinelearning/webnn#885 (6 comments by @anssiko, @fdwr, @Honry)
-
Future-proofing privacy considerations of New API feature to identify feature support per operator webmachinelearning/webnn#875 (1 comment by @anssiko)
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836 (1 comment by @zolkis)
-
-
4 pull requests opened.
-
Correct pad scalar rank inconsistency webmachinelearning/webnn#894, by @fdwr.
-
Add resample data type uint8/int8 webmachinelearning/webnn#891, by @fdwr.
-
Revise privacy considerations webmachinelearning/webnn#890, by @anssiko.
-
Rename cast/constant parameter `type` to `dataType` webmachinelearning/webnn#888, by @fdwr.
-
-
6 pull requests merged.
-
Correct pad scalar rank inconsistency webmachinelearning/webnn#894.
-
Add resample data type uint8/int8 webmachinelearning/webnn#891.
-
Support rankRange for op output tensors in opSupportLimits webmachinelearning/webnn#857.
-
Rename pool2d MLRoundingType - Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#770.
-
Rename cast/constant parameter `type` to `dataType` webmachinelearning/webnn#888.
-
Add opSupportLimits() privacy considerations for fingerprinting webmachinelearning/webnn#881.
-
-
4 pull requests received 31 new comments.
-
Support rankRange for op output tensors in opSupportLimits webmachinelearning/webnn#857 (20 comments by @fdwr, @huningxin)
-
Rename pool2d MLRoundingType - Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#770 (9 comments by @fdwr, @huningxin)
-
Add opSupportLimits() privacy considerations for fingerprinting webmachinelearning/webnn#881 (1 comment by @anssiko)
-
Update explainer with new proposal for simple accelerator mapping webmachinelearning/webnn#884 (1 comment by @zolkis)
-
TR publications
-
3 documents published.
-
Web Neural Network API published 3 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
2 participants joined.
-
Markus Tavenrath (NVIDIA) joined.
-
Joshua Lochner (Hugging Face) joined.
-
-
1 organization joined.
-
Hugging Face joined.
-
GitHub
webmachinelearning/webnn
-
7 issues opened.
-
Support flexible input sizes webmachinelearning/webnn#883, by @huningxin.
-
Propose to remove WPT conformance tests with negative scale of dequantizeLinear and quantizeLinear operator webmachinelearning/webnn#879, by @BruceDai.
-
How to support scatterElements on TFLite backend when indices operand is not constant webmachinelearning/webnn#878, by @fujunwei.
-
How to support edge padding mode on TFLite backend webmachinelearning/webnn#877, by @fujunwei.
-
How to support explicit padding on TFLite backend webmachinelearning/webnn#876, by @fujunwei.
-
Future-proofing privacy considerations of New API feature to identify feature support per operator webmachinelearning/webnn#875, by @sandandsnow.
-
Normalize the behavior when NaN is used for minValue or maxValue of clamp operator webmachinelearning/webnn#874, by @wangw-1991.
-
-
3 issues closed.
-
Clarify 0D tensor (rank of 0) support for axis-dependent operations webmachinelearning/webnn#873.
-
Behavior when there are NaNs in argmin/max inputs webmachinelearning/webnn#811.
-
Consider adding int64/uint64 data type support for some reduce operators webmachinelearning/webnn#694.
-
-
11 issues received 19 new comments.
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (6 comments by @handellm, @mwyrzykowski, @zolkis)
-
Support uint8/int8 input for resample2d webmachinelearning/webnn#872 (2 comments by @fdwr, @huningxin)
-
Clarify 0D tensor (rank of 0) support for axis-dependent operations webmachinelearning/webnn#873 (2 comments by @fdwr)
-
Consider adding int64/uint64 data type support for some reduce operators webmachinelearning/webnn#694 (1 comment by @anssiko)
-
Propose to remove WPT conformance tests with negative scale of dequantizeLinear and quantizeLinear operator webmachinelearning/webnn#879 (1 comment by @fdwr)
-
Normalize the behavior when NaN is used for minValue or maxValue of clamp operator webmachinelearning/webnn#874 (1 comment by @fdwr)
-
-
3 pull requests opened.
-
Update explainer with new proposal for simple accelerator mapping webmachinelearning/webnn#884, by @zolkis.
-
Bugfix: Only allow 1 to N rank input for operators that take 1 axis webmachinelearning/webnn#882, by @huningxin.
-
Add opSupportLimits() privacy considerations for fingerprinting webmachinelearning/webnn#881, by @anssiko.
-
-
3 pull requests merged.
-
Bugfix: Only allow 1 to N rank input for operators that take 1 axis webmachinelearning/webnn#882.
-
Add Accessibility Considerations webmachinelearning/webnn#869.
-
Add `isNaN` and `isInfinite` operators webmachinelearning/webnn#858.
-
-
5 pull requests received 12 new comments.
-
Add opSupportLimits() privacy considerations for fingerprinting webmachinelearning/webnn#881 (4 comments by @anssiko, @philloooo, @reillyeon)
-
Rename pool2d MLRoundingType - Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#770 (2 comments by @fdwr, @huningxin)
-
TR publications
-
1 document published.
-
Web Neural Network API published as a Candidate Recommendation Draft. History of Web Neural Network API
-
Participants
-
3 participants joined.
-
Johann Hofmann (Google LLC) joined.
-
Leo Lee (Microsoft Corporation) joined.
-
Ugur Acar (Depixen) joined.
-
-
1 participant left.
-
Joshua Lochner (Hugging Face) left.
-
-
1 organization joined.
-
Depixen joined.
-
GitHub
webmachinelearning/webnn
-
2 issues opened.
-
Clarify 0D tensor (rank of 0) support for axis-dependent operations webmachinelearning/webnn#873, by @miaobin.
-
Support uint8/int8 input for resample2d webmachinelearning/webnn#872, by @huningxin.
-
-
1 issue closed.
-
Rounding operators webmachinelearning/webnn#817.
-
-
1 issue reopened.
-
Request standards positions from Mozilla and WebKit webmachinelearning/webnn#763, by @reillyeon.
-
-
3 issues received 4 new comments.
-
Request standards positions from Mozilla and WebKit webmachinelearning/webnn#763 (2 comments by @iampi31415, @reillyeon)
-
Clarify 0D tensor (rank of 0) support for axis-dependent operations webmachinelearning/webnn#873 (1 comment by @miaobin)
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (1 comment by @mwyrzykowski)
-
-
1 pull request merged.
-
Add `roundEven` operator webmachinelearning/webnn#859.
-
-
2 pull requests received 10 new comments.
-
Rename pool2d MLRoundingType - Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#770 (8 comments by @dontcallmedom, @fdwr, @huningxin)
-
TR publications
-
3 documents published.
-
Web Neural Network API published 3 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
3 participants joined.
-
David Bokan (Google LLC) joined.
-
Hannah Van Opstal (Google LLC) joined.
-
Khushal Sagar (Google LLC) joined.
-
-
2 participants left.
-
Juan Deng (Alibaba Group) left.
-
Xiaojian Wang (Alibaba Group) left.
-
GitHub
webmachinelearning/webnn
-
6 issues opened.
-
Improve accessibility of examples webmachinelearning/webnn#870, by @anssiko.
-
May we update `reversed` options of `cumulativeSum` operation as `reverse`? webmachinelearning/webnn#867, by @BruceDai.
-
The output's allowed data types of Constraints table should be "uint8" for Element-wise logical operations webmachinelearning/webnn#865, by @BruceDai.
-
Typos of `input.shape` in dequantizeLinear decomposition code webmachinelearning/webnn#863, by @BruceDai.
-
Evaluate sustainability impact webmachinelearning/webnn#861, by @anssiko.
-
Consider drop the support for int32/uint32 of zeropoint for quantizeLinear webmachinelearning/webnn#856, by @lisa0314.
-
-
3 issues closed.
-
Consider removing or redesigning the `triangular` operator webmachinelearning/webnn#768.
-
The output's allowed data types of Constraints table should be "uint8" for Element-wise logical operations webmachinelearning/webnn#865.
-
Typos of `input.shape` in dequantizeLinear decomposition code webmachinelearning/webnn#863.
-
-
8 issues received 30 new comments.
-
Consider drop the support for int32/uint32 of zeropoint for quantizeLinear webmachinelearning/webnn#856 (8 comments by @fdwr, @huningxin, @philloooo, @reillyeon)
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836 (5 comments by @mwyrzykowski, @philloooo, @zolkis)
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (4 comments by @anssiko, @handellm)
-
Is there a reason to support out of order `axes` for layerNormalization? webmachinelearning/webnn#748 (2 comments by @anssiko, @fdwr)
-
Consider adding int64/uint64 data type support for some reduce operators webmachinelearning/webnn#694 (1 comment by @fdwr)
-
Consider removing or redesigning the `triangular` operator webmachinelearning/webnn#768 (1 comment by @anssiko)
-
-
11 pull requests opened.
-
Update WebNN architecture diagram SVG webmachinelearning/webnn#871, by @anssiko.
-
Add Accessibility Considerations webmachinelearning/webnn#869, by @anssiko.
-
Add a non-normative table of operators by category webmachinelearning/webnn#868, by @anssiko.
-
Bugfix: modify output's allowed data type as "uint8" for element-wise logical operations webmachinelearning/webnn#866, by @BruceDai.
-
Bugfix: Fix typos in dequantizeLinear's decomposition code webmachinelearning/webnn#864, by @BruceDai.
-
Add WebNN MLGraph Cache Explainer webmachinelearning/webnn#862, by @anssiko.
-
Update with an example HW selection guide and new use cases webmachinelearning/webnn#860, by @zolkis.
-
Add `roundEven` operator webmachinelearning/webnn#859, by @fdwr.
-
Add `isNaN` and `isInfinite` operators webmachinelearning/webnn#858, by @fdwr.
-
Support rankRange for op output tensors in opSupportLimits webmachinelearning/webnn#857, by @huningxin.
-
Add device preference use cases to explainer webmachinelearning/webnn#855, by @anssiko.
-
-
8 pull requests merged.
-
Update with an example HW selection guide and new use cases webmachinelearning/webnn#860.
-
Bugfix: Add missing 64-bit integers support for some reduction operators webmachinelearning/webnn#695.
-
Update WebNN architecture diagram SVG webmachinelearning/webnn#871.
-
Add a non-normative table of operators by category webmachinelearning/webnn#868.
-
Add WebNN MLGraph Cache Explainer webmachinelearning/webnn#862.
-
Bugfix: modify output's allowed data type as "uint8" for element-wise logical operations webmachinelearning/webnn#866.
-
Bugfix: Fix typos in dequantizeLinear's decomposition code webmachinelearning/webnn#864.
-
Add device preference use cases to explainer webmachinelearning/webnn#855.
-
-
11 pull requests received 105 new comments.
-
Add a non-normative table of operators by category webmachinelearning/webnn#868 (31 comments by @anssiko, @fdwr, @huningxin)
-
Add WebNN MLGraph Cache Explainer webmachinelearning/webnn#862 (26 comments by @anssiko, @Honry, @huningxin, @reillyeon)
-
Add device preference use cases to explainer webmachinelearning/webnn#855 (14 comments by @anssiko, @zolkis)
-
Update with an example HW selection guide and new use cases webmachinelearning/webnn#860 (12 comments by @anssiko, @handellm, @zolkis)
-
Update WebNN architecture diagram SVG webmachinelearning/webnn#871 (6 comments by @anssiko, @tomayac)
-
Add `isNaN` and `isInfinite` operators webmachinelearning/webnn#858 (6 comments by @fdwr, @huningxin)
-
Support rankRange for op output tensors in opSupportLimits webmachinelearning/webnn#857 (2 comments by @huningxin)
-
Bugfix: Add missing 64-bit integers support for some reduction operators webmachinelearning/webnn#695 (1 comment by @huningxin)
-
TR publications
-
4 documents published.
-
Web Neural Network API published 4 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
5 participants joined.
-
Sun Shin (NVIDIA) joined.
-
Suraj Bisht (Capital One Discover Network) joined.
-
Peter Tanski (Capital One Discover Network) joined.
-
Jonathan Schneerson (Temporal Series AI) joined.
-
Ehsan Toreini (Samsung Electronics Co., Ltd.) joined.
-
-
1 participant left.
-
Natasha Gaitonde (Microsoft Corporation) left.
-
-
2 organizations joined.
-
Capital One Discover Network joined.
-
Temporal Series AI joined.
-
GitHub
webmachinelearning/webnn
-
1 issue opened.
-
The minimum data type set webmachinelearning/webnn#853, by @huningxin.
-
-
5 issues closed.
-
String metadata and localization for operator labels webmachinelearning/webnn#837.
-
Clarify WebNN explainer and update diagram webmachinelearning/webnn#840.
-
The allowed data types of input operand for `sign` operator should also include `int64` type webmachinelearning/webnn#845.
-
Limited support for pad on CoreML backend webmachinelearning/webnn#739.
-
Need clarify constraint of 'beginningPadding' and 'endingPadding' for pad operation in "reflection" and "symmetric" mode webmachinelearning/webnn#377.
-
-
2 issues reopened.
-
Clarify WebNN explainer and update diagram webmachinelearning/webnn#840, by @anssiko.
-
String metadata and localization for operator labels webmachinelearning/webnn#837, by @inexorabletash.
-
-
10 issues received 23 new comments.
-
Clarify WebNN explainer and update diagram webmachinelearning/webnn#840 (6 comments by @anssiko, @huningxin, @iampi31415)
-
Consider removing or redesigning the `triangular` operator webmachinelearning/webnn#768 (4 comments by @fdwr, @huningxin, @xenova)
-
String metadata and localization for operator labels webmachinelearning/webnn#837 (3 comments by @anssiko, @xfq)
-
How to define the algorithm of L2_Pool2d? webmachinelearning/webnn#278 (2 comments by @fdwr, @reillyeon)
-
The allowed data types of input operand for `sign` operator should also include `int64` type webmachinelearning/webnn#845 (2 comments by @fdwr, @huningxin)
-
opSupportLimits: Level of detail for output tensor(s)? webmachinelearning/webnn#835 (1 comment by @inexorabletash)
-
Is there a reason to support out of order `axes` for layerNormalization? webmachinelearning/webnn#748 (1 comment by @fdwr)
-
-
8 pull requests opened.
-
define graph.devices webmachinelearning/webnn#854, by @philloooo.
-
Update README.md with Development guidance webmachinelearning/webnn#852, by @anssiko.
-
Note Unicode sequences in Security Considerations webmachinelearning/webnn#851, by @anssiko.
-
Apply clang-format to JS blocks via tools/reformat-js.py webmachinelearning/webnn#850, by @anssiko.
-
Clarify explainer based on review feedback webmachinelearning/webnn#849, by @anssiko.
-
Bugfix: Support `int64` for `abs`, `neg`, `sign`, `prelu` and `relu` webmachinelearning/webnn#848, by @huningxin.
-
Update WebNN architecture diagram webmachinelearning/webnn#847, by @anssiko.
-
Update several docs repo links webmachinelearning/webnn#846, by @guspan-tanadi.
-
-
10 pull requests merged.
-
Update README.md with Development guidance webmachinelearning/webnn#852.
-
Apply clang-format to JS blocks via tools/reformat-js.py webmachinelearning/webnn#850.
-
Note Unicode sequences in Security Considerations webmachinelearning/webnn#851.
-
Clarify explainer based on review feedback webmachinelearning/webnn#849.
-
Update WebNN architecture diagram webmachinelearning/webnn#847.
-
Bugfix: Support `int64` for `abs`, `neg`, `sign`, `prelu` and `relu` webmachinelearning/webnn#848.
-
Update several docs repo links webmachinelearning/webnn#846.
-
Add notes regarding label usage, provided by i18n review webmachinelearning/webnn#841.
-
MLTensor explainer: replace "import buffer" with "export tensor" webmachinelearning/webnn#844.
-
Restrict padding options to better match backend limits webmachinelearning/webnn#843.
-
-
8 pull requests received 45 new comments.
-
Apply clang-format to JS blocks via tools/reformat-js.py webmachinelearning/webnn#850 (13 comments by @anssiko, @fdwr, @huningxin, @inexorabletash)
-
Clarify explainer based on review feedback webmachinelearning/webnn#849 (8 comments by @anssiko, @huningxin, @reillyeon)
-
Update WebNN architecture diagram webmachinelearning/webnn#847 (7 comments by @anssiko, @huningxin, @inexorabletash, @reillyeon)
-
Add notes regarding label usage, provided by i18n review webmachinelearning/webnn#841 (6 comments by @anssiko, @inexorabletash, @xfq)
-
Update README.md with Development guidance webmachinelearning/webnn#852 (4 comments by @anssiko, @fdwr, @huningxin)
-
Update several docs repo links webmachinelearning/webnn#846 (3 comments by @anssiko, @guspan-tanadi, @w3cbot)
-
Bugfix: Add missing 64-bit integers support for some reduction operators webmachinelearning/webnn#695 (2 comments by @huningxin)
-
Bugfix: Support `int64` for `abs`, `neg`, `sign`, `prelu` and `relu` webmachinelearning/webnn#848 (2 comments by @huningxin)
-
TR publications
-
4 documents published.
-
Web Neural Network API published 4 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
7 participants joined.
-
Taylore Givens (Microsoft Corporation) joined.
-
Kevin Petit (ARM Limited) joined.
-
Eugen Thaci (CleverThis, Inc.) joined.
-
Josi Rosenfeld (CleverThis, Inc.) joined.
-
Adnaan Nazir (CleverThis, Inc.) joined.
-
Drew Morris (CleverThis, Inc.) joined.
-
Aditya Chhabra (CleverThis, Inc.) joined.
-
-
1 participant left.
-
Margery Clacy (Centre for Inclusive Design) left.
-
-
1 organization joined.
-
CleverThis, Inc. joined.
-
-
1 organization left.
-
Centre for Inclusive Design left.
-
GitHub
webmachinelearning/webnn
-
5 issues opened.
-
The allowed data types of input operand for `sign` operator should also include `int64` type webmachinelearning/webnn#845, by @BruceDai.
-
Clarify WebNN explainer and update diagram webmachinelearning/webnn#840, by @iampi31415.
-
String metadata and localization for operator labels webmachinelearning/webnn#837, by @xfq.
-
Get devices used for a graph after graph compilation webmachinelearning/webnn#836, by @philloooo.
-
opSupportLimits: Level of detail for output tensor(s)? webmachinelearning/webnn#835, by @inexorabletash.
-
-
2 issues closed.
-
Support building graphs from `MLTensor` containing constants webmachinelearning/webnn#760.
-
Define the maximum number of operand dimensions (maximum rank) webmachinelearning/webnn#456.
-
-
10 issues received 47 new comments.
-
Caching mechanism for MLGraph webmachinelearning/webnn#807 (12 comments by @anssiko, @KenjiBaheux, @mmccool, @reillyeon)
-
Limited support for pad on CoreML backend webmachinelearning/webnn#739 (7 comments by @fdwr, @huningxin, @xenova)
-
Clarify WebNN explainer and update diagram webmachinelearning/webnn#840 (6 comments by @iampi31415, @inexorabletash, @reillyeon)
-
Request standards positions from Mozilla and WebKit webmachinelearning/webnn#763 (4 comments by @iampi31415, @inexorabletash, @reillyeon)
-
String metadata and localization for operator labels webmachinelearning/webnn#837 (3 comments by @inexorabletash, @xfq)
-
opSupportLimits: Level of detail for output tensor(s)? webmachinelearning/webnn#835 (3 comments by @inexorabletash, @philloooo)
-
Need clarify constraint of 'beginningPadding' and 'endingPadding' for pad operation in "reflection" and "symmetric" mode webmachinelearning/webnn#377 (2 comments by @fdwr, @huningxin)
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (2 comments by @anssiko)
-
-
7 pull requests opened.
-
MLTensor explainer: replace "import buffer" with "export tensor" webmachinelearning/webnn#844, by @bbernhar.
-
Restrict padding options to better match backend limits webmachinelearning/webnn#843, by @inexorabletash.
-
Editorial: Improve some wording around op creation steps webmachinelearning/webnn#842, by @inexorabletash.
-
Add notes regarding label usage, provided by i18n review webmachinelearning/webnn#841, by @inexorabletash.
-
Editorial: Fix some option links and one typo webmachinelearning/webnn#839, by @inexorabletash.
-
Tooling: Better documentation for how linting works webmachinelearning/webnn#838, by @inexorabletash.
-
Editorial: Miscellaneous cleanups webmachinelearning/webnn#834, by @inexorabletash.
-
-
6 pull requests merged.
-
Editorial: Improve some wording around op creation steps webmachinelearning/webnn#842.
-
Editorial: Fix some option links and one typo webmachinelearning/webnn#839.
-
Tooling: Better documentation for how linting works webmachinelearning/webnn#838.
-
Allow tensors for graph constants. webmachinelearning/webnn#830.
-
Add rankRange to opSupportLimits() webmachinelearning/webnn#828.
-
Editorial: Miscellaneous cleanups webmachinelearning/webnn#834.
-
-
9 pull requests received 64 new comments.
-
Allow tensors for graph constants. webmachinelearning/webnn#830 (24 comments by @bbernhar, @fdwr, @huningxin, @inexorabletash, @reillyeon)
-
Add rankRange to opSupportLimits() webmachinelearning/webnn#828 (13 comments by @fdwr, @huningxin, @inexorabletash, @philloooo, @reillyeon)
-
Editorial: Miscellaneous cleanups webmachinelearning/webnn#834 (7 comments by @anssiko, @fdwr, @huningxin, @inexorabletash)
-
Tooling: Better documentation for how linting works webmachinelearning/webnn#838 (6 comments by @fdwr, @huningxin, @inexorabletash)
-
Restrict padding options to better match backend limits webmachinelearning/webnn#843 (5 comments by @fdwr, @huningxin, @inexorabletash)
-
MLTensor explainer: replace "import buffer" with "export tensor" webmachinelearning/webnn#844 (4 comments by @bbernhar, @reillyeon)
-
Editorial: Improve some wording around op creation steps webmachinelearning/webnn#842 (2 comments by @huningxin, @inexorabletash)
-
Editorial: Fix some option links and one typo webmachinelearning/webnn#839 (2 comments by @huningxin, @inexorabletash)
-
Add notes regarding label usage, provided by i18n review webmachinelearning/webnn#841 (1 comment by @inexorabletash)
-
Lifecycle
-
1 charter started.
-
Chartered until 30 April 2027.
-
TR publications
-
3 documents published.
-
Web Neural Network API published 3 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
4 participants joined.
-
Elena Zhelezina (ARM Limited) joined.
-
Matthew Atkinson (Samsung Electronics Co., Ltd.) joined.
-
Winston Chen (Samsung Electronics Co., Ltd.) joined.
-
Brent Zundel (Tradeverifyd) joined.
-
-
1 participant left.
-
Warren Huffman (Tradeverifyd) left.
-
-
1 organization joined.
-
ARM Limited joined.
-
GitHub
webmachinelearning/webnn
-
4 issues opened.
-
Consider drop symmetric mode for Pad webmachinelearning/webnn#833, by @lisa0314.
-
Consider only support consecutive axes for LayerNormalization webmachinelearning/webnn#832, by @lisa0314.
-
Strong typing for readTensor() and writeTensor() webmachinelearning/webnn#826, by @reillyeon.
-
Always allow Uint8Array in constant() as a generic ArrayBuffer slice type webmachinelearning/webnn#825, by @reillyeon.
-
-
15 issues closed.
-
[MLBuffer] Creation and representing MLBuffer on a XPU devices webmachinelearning/webnn#542.
-
Consider drop symmetric mode for Pad webmachinelearning/webnn#833.
-
Consider only support consecutive axes for LayerNormalization webmachinelearning/webnn#832.
-
Consider dropping the support of uint64/int64 data type for some operators webmachinelearning/webnn#654.
-
Proposal: Report non-fatal errors from the WebNN timeline webmachinelearning/webnn#778.
-
Always allow Uint8Array in constant() as a generic ArrayBuffer slice type webmachinelearning/webnn#825.
-
Clarify at which point weights are used in the compilation webmachinelearning/webnn#157.
-
Support for transformers webmachinelearning/webnn#375.
-
Strong typing for readTensor() and writeTensor() webmachinelearning/webnn#826.
-
Request the decomposition for gatherElements, scatterElements and scatterND webmachinelearning/webnn#767.
-
Support block-wise quantization webmachinelearning/webnn#779.
-
Support strides option for `slice` operator webmachinelearning/webnn#772.
-
Support `reverse` operator webmachinelearning/webnn#773.
-
Where is scatter and gather op? webmachinelearning/webnn#467.
-
Add QuantizeLinear and DequantizeLinear for mixed precision webmachinelearning/webnn#93.
-
-
1 issue reopened.
-
Wide review tracker webmachinelearning/webnn#239, by @anssiko.
-
-
19 issues received 76 new comments.
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (16 comments by @anssiko, @handellm, @inexorabletash, @mwyrzykowski, @reillyeon, @zolkis)
-
(de)quantization behaviors on CoreML webmachinelearning/webnn#822 (12 comments by @fdwr, @huningxin, @philloooo)
-
Caching mechanism for MLGraph webmachinelearning/webnn#807 (12 comments by @anssiko, @inexorabletash, @mmccool, @reillyeon, @zolkis)
-
Expose WebNN API to service workers webmachinelearning/webnn#804 (5 comments by @anssiko, @inexorabletash, @zweack)
-
Limited support for pad on CoreML backend webmachinelearning/webnn#739 (4 comments by @fdwr, @huningxin, @lisa0314)
-
Consider only support consecutive axes for LayerNormalization webmachinelearning/webnn#832 (4 comments by @huningxin, @inexorabletash, @lisa0314)
-
Consider drop symmetric mode for Pad webmachinelearning/webnn#833 (3 comments by @huningxin, @inexorabletash, @lisa0314)
-
Request the decomposition for gatherElements, scatterElements and scatterND webmachinelearning/webnn#767 (3 comments by @fdwr, @inexorabletash)
-
Need clarify constraint of 'beginningPadding' and 'endingPadding' for pad operation in "reflection" and "symmetric" mode webmachinelearning/webnn#377 (2 comments by @fdwr, @huningxin)
-
Consider dropping the support of uint64/int64 data type for some operators webmachinelearning/webnn#654 (2 comments by @inexorabletash, @lisa0314)
-
Strong typing for readTensor() and writeTensor() webmachinelearning/webnn#826 (2 comments by @inexorabletash, @reillyeon)
-
[MLBuffer] Creation and representing MLBuffer on a XPU devices webmachinelearning/webnn#542 (1 comment by @bbernhar)
-
Is there a reason to support out of order `axes` for layerNormalization? webmachinelearning/webnn#748 (1 comment by @lisa0314)
-
Proposal: Report non-fatal errors from the WebNN timeline webmachinelearning/webnn#778 (1 comment by @a-sully)
-
webmachinelearning/webnn#157 (1 comment by @RafaelCintron)
-
Always allow Uint8Array in constant() as a generic ArrayBuffer slice type webmachinelearning/webnn#825 (1 comment by @inexorabletash)
-
-
5 pull requests opened.
-
Tooling: Use WebIDL2 parser when linting to identify dictionaries webmachinelearning/webnn#831, by @inexorabletash.
-
Allow tensors for graph constants. webmachinelearning/webnn#830, by @bbernhar.
-
Allow Uint8Array in constant() etc. as a generic ArrayBufferView type webmachinelearning/webnn#829, by @inexorabletash.
-
Add rankRange to opSupportLimits() webmachinelearning/webnn#828, by @inexorabletash.
-
Lint: Add additional checks and improve HTML parsing webmachinelearning/webnn#827, by @inexorabletash.
-
-
6 pull requests merged.
-
Tooling: Use WebIDL2 parser when linting to identify dictionaries webmachinelearning/webnn#831.
-
Reject promises for in-progress operations on an MLTensor if it is destroyed webmachinelearning/webnn#799.
-
Allow Uint8Array in constant() etc. as a generic ArrayBufferView type webmachinelearning/webnn#829.
-
Lint: Add additional checks and improve HTML parsing webmachinelearning/webnn#827.
-
Operator set wave 3 webmachinelearning/webnn#805.
-
Update device selection explainer webmachinelearning/webnn#824.
-
-
8 pull requests received 199 new comments.
-
Operator set wave 3 webmachinelearning/webnn#805 (91 comments by @fdwr, @huningxin, @inexorabletash)
-
Allow tensors for graph constants. webmachinelearning/webnn#830 (68 comments by @anssiko, @bbernhar, @fdwr, @huningxin, @inexorabletash, @reillyeon)
-
Add rankRange to opSupportLimits() webmachinelearning/webnn#828 (26 comments by @fdwr, @huningxin, @inexorabletash, @philloooo, @reillyeon)
-
Lint: Add additional checks and improve HTML parsing webmachinelearning/webnn#827 (5 comments by @inexorabletash)
-
Allow Uint8Array in constant() etc. as a generic ArrayBufferView type webmachinelearning/webnn#829 (4 comments by @fdwr, @inexorabletash)
-
Tooling: Use WebIDL2 parser when linting to identify dictionaries webmachinelearning/webnn#831 (1 comment by @inexorabletash)
-
Reject promises for in-progress operations on an MLTensor if it is destroyed webmachinelearning/webnn#799 (1 comment by @a-sully)
-
webmachinelearning/webmachinelearning-ethics
-
1 issue opened.
-
International AI Safety Report webmachinelearning/webmachinelearning-ethics#33, by @anssiko.
-
-
1 issue received 3 new comments.
-
International AI Safety Report webmachinelearning/webmachinelearning-ethics#33 (3 comments by @anssiko, @chrisn, @dontcallmedom)
-
TR publications
-
5 documents published.
-
Web Neural Network API published 5 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
1 participant joined.
-
Brad Triebwasser (Google LLC) joined.
-
GitHub
webmachinelearning/webnn
-
5 issues opened.
-
(de)quantization behaviors on CoreML webmachinelearning/webnn#822, by @philloooo.
-
Operator naming 2D vs 2d webmachinelearning/webnn#821, by @fdwr.
-
Validate user-supplied `outputSizes` of `MLConvTranspose2dOptions` and `MLPool2dOptions` webmachinelearning/webnn#818, by @huningxin.
-
Rounding operators webmachinelearning/webnn#817, by @fdwr.
-
Query supported devices before graph compilation webmachinelearning/webnn#815, by @anssiko.
-
-
11 issues closed.
-
Expose WebNN API to service workers webmachinelearning/webnn#804.
-
Behavior for numeric overflows/underflows/etc in algorithms webmachinelearning/webnn#636.
-
Validate user-supplied `outputSizes` of `MLConvTranspose2dOptions` and `MLPool2dOptions` webmachinelearning/webnn#818.
-
Define equality for non-trivial data types webmachinelearning/webnn#792.
-
Support coordinate transformation modes for Resample2d webmachinelearning/webnn#270.
-
Inconsistency between MLGraphBuilder and MLBuffer construction webmachinelearning/webnn#697.
-
Rename destroy() to dispose() webmachinelearning/webnn#756.
-
Please clarify interpolation algorithm for resample2d webmachinelearning/webnn#358.
-
Need to understand how WebNN supports implementation that involves multiple devices and timelines webmachinelearning/webnn#350.
-
MLContextOptions.deviceType seems unnecessary outside of conformance testing webmachinelearning/webnn#749.
-
API simplification: context types, context options, createContext() webmachinelearning/webnn#302.
-
-
14 issues received 44 new comments.
-
Query supported devices before graph compilation webmachinelearning/webnn#815 (24 comments by @anssiko, @fsolenberg, @handellm, @huningxin, @inexorabletash, @mwyrzykowski, @reillyeon, @zolkis)
-
Support coordinate transformation modes for Resample2d webmachinelearning/webnn#270 (4 comments by @fdwr, @Honry, @inexorabletash)
-
Behavior when there are NaNs in argmin/max inputs webmachinelearning/webnn#811 (4 comments by @fdwr, @philloooo)
-
MLContextOptions.deviceType seems unnecessary outside of conformance testing webmachinelearning/webnn#749 (2 comments by @anssiko, @mwyrzykowski)
-
Request the decomposition for gatherElements, scatterElements and scatterND webmachinelearning/webnn#767 (1 comment by @fdwr)
-
Behavior for numeric overflows/underflows/etc in algorithms webmachinelearning/webnn#636 (1 comment by @inexorabletash)
-
Validate user-supplied `outputSizes` of `MLConvTranspose2dOptions` and `MLPool2dOptions` webmachinelearning/webnn#818 (1 comment by @inexorabletash)
-
Inconsistency between MLGraphBuilder and MLBuffer construction webmachinelearning/webnn#697 (1 comment by @inexorabletash)
-
API simplification: context types, context options, createContext() webmachinelearning/webnn#302 (1 comment by @zolkis)
-
-
7 pull requests opened.
-
Update device selection explainer webmachinelearning/webnn#824, by @zolkis.
-
Expose WebNN API to shared workers and service workers webmachinelearning/webnn#823, by @inexorabletash.
-
Add missing validation steps for several ops webmachinelearning/webnn#820, by @inexorabletash.
-
Fix returning failure from methods, and use "item" for lists webmachinelearning/webnn#819, by @inexorabletash.
-
Clarify interpolation algorithms for resample2d webmachinelearning/webnn#816, by @inexorabletash.
-
Editorial: Inline shape calculation algorithms only used once webmachinelearning/webnn#814, by @inexorabletash.
-
Define equality for lists and MLOperandDescriptors webmachinelearning/webnn#813, by @inexorabletash.
-
-
7 pull requests merged.
-
Expose WebNN API to shared workers and service workers webmachinelearning/webnn#823.
-
Add missing validation steps for several ops webmachinelearning/webnn#820.
-
Fix returning failure from methods, and use "item" for lists webmachinelearning/webnn#819.
-
Define equality for lists and MLOperandDescriptors webmachinelearning/webnn#813.
-
Editorial: Inline shape calculation algorithms only used once webmachinelearning/webnn#814.
-
Clarify interpolation algorithms for resample2d webmachinelearning/webnn#816.
-
Remove MLDeviceType webmachinelearning/webnn#809.
-
-
10 pull requests received 185 new comments.
-
Clarify interpolation algorithms for resample2d webmachinelearning/webnn#816 (29 comments by @fdwr, @huningxin, @inexorabletash)
-
Remove MLDeviceType webmachinelearning/webnn#809 (27 comments by @anssiko, @fdwr, @huningxin, @inexorabletash, @mwyrzykowski, @zolkis)
-
Add missing validation steps for several ops webmachinelearning/webnn#820 (12 comments by @huningxin, @inexorabletash)
-
Editorial: Inline shape calculation algorithms only used once webmachinelearning/webnn#814 (7 comments by @fdwr, @huningxin, @inexorabletash)
-
Update device selection explainer webmachinelearning/webnn#824 (3 comments by @anssiko, @inexorabletash, @zolkis)
-
Expose WebNN API to shared workers and service workers webmachinelearning/webnn#823 (2 comments by @inexorabletash)
-
Rename pool2d MLRoundingType - Simplify the operand layout support of conv2d and pooling 2d operations webmachinelearning/webnn#770 (2 comments by @inexorabletash)
-
Fix returning failure from methods, and use "item" for lists webmachinelearning/webnn#819 (1 comment by @inexorabletash)
-
Define equality for lists and MLOperandDescriptors webmachinelearning/webnn#813 (1 comment by @inexorabletash)
TR publications
-
6 documents published.
-
Web Neural Network API published 6 times as a Candidate Recommendation Draft. History of Web Neural Network API
-
Past meetings
Participants
-
4 participants joined.
-
Christine Hollingsworth (Google LLC) joined.
-
Mike Wasserman (Google LLC) joined.
-
Michał Pierzchała (Callstack) joined.
-
Michal Grabowski (Callstack) joined.
-
-
1 organization joined.
-
Callstack joined.
-
GitHub
webmachinelearning/webnn
-
4 issues opened.
-
Behavior when there are NaNs in argmin/max inputs webmachinelearning/webnn#811, by @philloooo.
-
Caching mechanism for MLGraph webmachinelearning/webnn#807, by @anssiko.
-
Expose WebNN API to service workers webmachinelearning/webnn#804, by @zweack.
-
Need reverse output sequence for backward direction of gru and lstm operator emulations webmachinelearning/webnn#801, by @BruceDai.
-
-
7 issues closed.
-
Proposed style/wording tweaks webmachinelearning/webnn#783.
-
Create new objects with a Realm webmachinelearning/webnn#793.
-
WebNN and StableHLO opset compatibility webmachinelearning/webnn#346.
-
[MLBuffer] Support for MLBuffer in graph execution webmachinelearning/webnn#544.
-
Need reverse output sequence for backward direction of gru and lstm operator emulations webmachinelearning/webnn#801.
-
[MLBuffer] Uploading/downloading tensor data webmachinelearning/webnn#543.
-
Allow NaN and infinite values for float attributes webmachinelearning/webnn#800.
-
-
15 issues received 43 new comments.
-
Caching mechanism for MLGraph webmachinelearning/webnn#807 (10 comments by @bbernhar, @inexorabletash, @mmccool, @reillyeon, @zolkis)
-
MLBuffer destruction timeline webmachinelearning/webnn#716 (4 comments by @a-sully, @bbernhar, @inexorabletash, @reillyeon)
-
Rename destroy() to dispose() webmachinelearning/webnn#756 (4 comments by @anssiko, @inexorabletash, @reillyeon)
-
Disallow operations on scalar tensors that are no-ops webmachinelearning/webnn#794 (4 comments by @fdwr, @huningxin)
-
Support building graphs from `MLTensor` containing constants webmachinelearning/webnn#760 (3 comments by @a-sully, @bbernhar)
-
[MLBuffer] Support for MLBuffer in graph execution webmachinelearning/webnn#544 (3 comments by @bbernhar, @inexorabletash)
-
Need reverse output sequence for backward direction of gru and lstm operator emulations webmachinelearning/webnn#801 (3 comments by @BruceDai, @inexorabletash, @philloooo)
-
[MLBuffer] Uploading/downloading tensor data webmachinelearning/webnn#543 (3 comments by @bbernhar, @fdwr, @inexorabletash)
-
Inconsistency between MLGraphBuilder and MLBuffer construction webmachinelearning/webnn#697 (2 comments by @inexorabletash, @reillyeon)
-
Need to understand how WebNN supports implementation that involves multiple devices and timelines webmachinelearning/webnn#350 (1 comment by @plehegar)
-
Allow NaN and infinite values for float attributes webmachinelearning/webnn#800 (1 comment by @huningxin)
-
-
7 pull requests opened.
-
Ensure object creation specifies the realm webmachinelearning/webnn#810, by @inexorabletash.
-
Remove MLDeviceType webmachinelearning/webnn#809, by @zolkis.
-
update privacy readme for opSupportLimits webmachinelearning/webnn#808, by @philloooo.
-
Editorial: Fix glitch with otherwise clause in pool op creation webmachinelearning/webnn#806, by @inexorabletash.
-
Operator set wave 3 webmachinelearning/webnn#805, by @fdwr.
-
Bugfix: Fix emulation error of gru by 'backward' and 'both' direction options webmachinelearning/webnn#803, by @BruceDai.
-
Bugfix: Fix emulation error of lstm by 'backward' and 'both' direction options webmachinelearning/webnn#802, by @BruceDai.
-
-
8 pull requests merged.
-
Editorial: Various style and wording tweaks webmachinelearning/webnn#797.
-
Ensure object creation specifies the realm webmachinelearning/webnn#810.
-
Editorial: Fix glitch with otherwise clause in pool op creation webmachinelearning/webnn#806.
-
update privacy readme for opSupportLimits webmachinelearning/webnn#808.
-
Bugfix: Fix emulation error of gru by 'backward' and 'both' direction options webmachinelearning/webnn#803.
-
Bugfix: Fix emulation error of lstm by 'backward' and 'both' direction options webmachinelearning/webnn#802.
-
Add device selection explainer webmachinelearning/webnn#784.
-
Specify destroy() methods and behavior around context loss and error reporting webmachinelearning/webnn#744.
-
-
11 pull requests received 143 new comments.
-
Operator set wave 3 webmachinelearning/webnn#805 (41 comments by @anssiko, @fdwr, @huningxin, @inexorabletash)
-
Specify destroy() methods and behavior around context loss and error reporting webmachinelearning/webnn#744 (28 comments by @huningxin, @inexorabletash, @mingmingtasd, @mwyrzykowski, @reillyeon)
-
Add device selection explainer webmachinelearning/webnn#784 (26 comments by @anssiko, @fdwr, @zolkis)
-
Remove MLDeviceType webmachinelearning/webnn#809 (16 comments by @anssiko, @fdwr, @inexorabletash, @zolkis)
-
Ensure object creation specifies the realm webmachinelearning/webnn#810 (8 comments by @fdwr, @inexorabletash)
-
Editorial: Fix glitch with otherwise clause in pool op creation webmachinelearning/webnn#806 (7 comments by @fdwr, @huningxin, @inexorabletash)
-
Editorial: Various style and wording tweaks webmachinelearning/webnn#797 (5 comments by @fdwr, @inexorabletash)
-
Bugfix: Fix emulation error of lstm by 'backward' and 'both' direction options webmachinelearning/webnn#802 (4 comments by @BruceDai, @huningxin, @philloooo)
-
Bugfix: Fix emulation error of gru by 'backward' and 'both' direction options webmachinelearning/webnn#803 (4 comments by @BruceDai, @huningxin, @philloooo)
-
update privacy readme for opSupportLimits webmachinelearning/webnn#808 (3 comments by @fdwr, @inexorabletash, @philloooo)
-
Reject promises for in-progress operations on an MLTensor if it is destroyed webmachinelearning/webnn#799 (1 comment by @fdwr)
-