{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":529966861,"defaultBranch":"main","name":"onnxruntime","ownerLogin":"titaiwangms","currentUserCanPush":false,"isFork":true,"isEmpty":false,"createdAt":"2022-08-28T20:03:00.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/18010845?v=4","public":true,"private":false,"isOrgOwned":false},"refInfo":{"name":"","listCacheKey":"v0:1691511971.0","currentOid":""},"activityList":{"items":[{"before":"ad382120fe4b9b17d7dbb871feb397e8dfd183af","after":"af18824f430225330ca58dd9a5a08bdc57900955","ref":"refs/heads/main","pushedAt":"2024-08-26T23:34:21.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[JS/WebGPU] Add GatherBlockQuantized op support (#21734)\n\n### Description\r\nAdd GatherBlockQuantized operator to JSEP.\r\n\r\n\r\n\r\n### Motivation and Context\r\nGemma model requires this.","shortMessageHtmlLink":"[JS/WebGPU] Add GatherBlockQuantized op support (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2464954092\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21734\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21734/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21734\">microsoft#21734</a>)"}},{"before":"d9c57ac7db84f56a1e47e595f59bd792aa44de57","after":"ad382120fe4b9b17d7dbb871feb397e8dfd183af","ref":"refs/heads/main","pushedAt":"2024-08-26T21:37:22.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[CUDA] enable causal in MultiHeadAttention (#21852)\n\n### Description\r\nEnable causal in MultiHeadAttention cuda operator.\r\n\r\nAll formats (Q_K_V_BSNH_BSNH_BSNH, Q_K_V_BSNH_BNSH_BNSH, Q_KV_BSNH_BSN2H\r\nand QKV_BSN3H) supports causal for now. Internally, casual will be\r\ndispatch to flash attention, efficient attention or unfused attention\r\nkernel.\r\n\r\n### Motivation and Context\r\nCurrently, MultiHeadAttention has causal enabled in CPU ep, but not in\r\nCUDA ep. It could cause issues in onnx conversion, like some model can\r\nrun in CPU but not in CUDA. Enable causal in CUDA will reduce the\r\ndifference of support matrix of CPU/CUDA.","shortMessageHtmlLink":"[CUDA] enable causal in MultiHeadAttention (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2485397089\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21852\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21852/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21852\">microsoft#21852</a>)"}},{"before":"c5592fdcef55b833c99e9e943ef0112100d51375","after":"d9c57ac7db84f56a1e47e595f59bd792aa44de57","ref":"refs/heads/main","pushedAt":"2024-08-26T19:08:22.000Z","pushType":"push","commitsCount":66,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[js/webgpu] Enable pad f16 uniform (#21691)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->\r\n\r\n---------\r\n\r\nCo-authored-by: Yulong Wang <7679871+fs-eire@users.noreply.github.com>","shortMessageHtmlLink":"[js/webgpu] Enable pad f16 uniform (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2457659355\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21691\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21691/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21691\">microsoft#21691</a>)"}},{"before":"d0a6f57d74a62be07314780aa061d138200936d0","after":"c5592fdcef55b833c99e9e943ef0112100d51375","ref":"refs/heads/main","pushedAt":"2024-08-12T22:40:27.000Z","pushType":"push","commitsCount":41,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[DML EP] Update DML to 1.15.1 (#21695)\n\n### Description\r\nUpdate DML runtime binary to 1.15.1\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"[DML EP] Update DML to 1.15.1 (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2458795370\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21695\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21695/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21695\">microsoft#21695</a>)"}},{"before":"1637f22d39b6d57d2774d7d41e6a8ae1815180c5","after":"d0a6f57d74a62be07314780aa061d138200936d0","ref":"refs/heads/main","pushedAt":"2024-08-02T16:43:52.000Z","pushType":"push","commitsCount":11,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Add reduce kernels for bigger types (#21490)","shortMessageHtmlLink":"Add reduce kernels for bigger types (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2428754687\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21490\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21490/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21490\">microsoft#21490</a>)"}},{"before":"adaf0e81168a32040839e3bfea09ac69edcb011e","after":"1637f22d39b6d57d2774d7d41e6a8ae1815180c5","ref":"refs/heads/main","pushedAt":"2024-07-30T18:51:01.000Z","pushType":"push","commitsCount":160,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Extend Pad Fusion for AveragePool (#21556)\n\n### Description\r\nThis extends the existing pad_fusion for AveragePool operator i.e. fuse\r\nPad if it is followed by AveragePool operator.\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"Extend Pad Fusion for AveragePool (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2436608332\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21556\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21556/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21556\">microsoft#21556</a>)"}},{"before":"c501c6ffaf214c71c49bc36e99f225e55e74925a","after":"adaf0e81168a32040839e3bfea09ac69edcb011e","ref":"refs/heads/main","pushedAt":"2024-06-24T19:53:39.000Z","pushType":"push","commitsCount":50,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[Fix] USE_NCCL -> ORT_USE_NCCL (#21136)\n\n### Description\r\nCorrect the macro used when NCCL enabled.","shortMessageHtmlLink":"[Fix] USE_NCCL -&gt; ORT_USE_NCCL (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2366004172\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21136\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21136/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21136\">microsoft#21136</a>)"}},{"before":"05032e5e5fff4d55b0178f27fba42161c255f3f5","after":"c501c6ffaf214c71c49bc36e99f225e55e74925a","ref":"refs/heads/main","pushedAt":"2024-06-17T16:26:20.000Z","pushType":"push","commitsCount":32,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Rename a mispelled filename in the documentation (#21066)\n\n### Description\r\nRename a file in the documentation","shortMessageHtmlLink":"Rename a mispelled filename in the documentation (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2356969611\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/21066\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/21066/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/21066\">microsoft#21066</a>)"}},{"before":"d13cabf7f9350c296373bf2f1d59d2ebd84f71de","after":"05032e5e5fff4d55b0178f27fba42161c255f3f5","ref":"refs/heads/main","pushedAt":"2024-06-11T18:04:24.000Z","pushType":"push","commitsCount":41,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Updating cudnn from 8 to 9 on exsiting cuda 12 docker image (#20925)\n\n### Description\r\nAdding support of cudnn 9 \r\n\r\n\r\n### Motivation and Context\r\nKeep exsiting  cuda 12.2 with nvidia dirver 535","shortMessageHtmlLink":"Updating cudnn from 8 to 9 on exsiting cuda 12 docker image (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2334525716\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20925\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20925/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20925\">microsof…</a>"}},{"before":"798cea2350a196a67ff7e0621ea125c7f2035f7c","after":"d13cabf7f9350c296373bf2f1d59d2ebd84f71de","ref":"refs/heads/main","pushedAt":"2024-06-03T18:53:29.000Z","pushType":"push","commitsCount":18,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Upgrade GCC and remove the dependency on GCC8's experimental std::filesystem implementation (#20893)\n\n### Description\r\nThis PR upgrades CUDA 11 build pipelines' GCC version from 8 to 11. \r\n\r\n### Motivation and Context\r\n\r\nGCC8 has an experimental std::filesystem implementation which is not ABI\r\ncompatible with the formal one in later GCC releases. It didn't cause\r\ntrouble for us, however, ONNX community has encountered this issue much.\r\nFor example, https://github.com/onnx/onnx/issues/6047 . So this PR\r\nincreases the minimum supported GCC version from 8 to 9, and removes the\r\nreferences to GCC's \"stdc++fs\" library. Please note we compile our code\r\non RHEL8 and RHEL8's libstdc++ doesn't have the fs library, which means\r\nthe binaries in ONNX Runtime's official packages always static link to\r\nthe fs library. It is just a matter of which version of the library, an\r\nexperimental one or a more mature one. And it is an implementation\r\ndetail that is not visible from outside. Anyway, a newer GCC is better.\r\nIt will give us the chance to use many C++20 features.\r\n\r\n#### Why we were using GCC 8?\r\nIt is because all our Linux packages were built on RHEL8 or its\r\nequivalents. The default GCC version in RHEL8 is 8. RHEL also provides\r\nadditional GCC versions from RH devtoolset. UBI8 is the abbreviation of\r\nRed Hat Universal Base Image 8, which is the containerized RHEL8. UBI8\r\nis free, which means it doesn't require a subscription(while RHEL does).\r\nThe only devtoolset that UBI8 provides is GCC 12, which is too new for\r\nbeing used with CUDA 11.8. And our CUDA 11.8's build env is a docker\r\nimage from Nvidia that is based on UBI8.\r\n#### How the problem is solved\r\nAlmalinux is an alternative to RHEL. Almalinux 8 provides GCC 11. And\r\nthe CUDA 11.8 docker image from Nvidia is open source, which means we\r\ncan rebuild the image based on Almalinux 8 to get GCC 11. I've done\r\nthis, but I cannot republish the new image due to various complicated\r\nlicense restrictions. Therefore I put them at an internal location in\r\nonnxruntimebuildcache.azurecr.io.","shortMessageHtmlLink":"Upgrade GCC and remove the dependency on GCC8's experimental std::fil…"}},{"before":"362a62390545234d28661307c7a203f2546280ae","after":"798cea2350a196a67ff7e0621ea125c7f2035f7c","ref":"refs/heads/main","pushedAt":"2024-05-29T17:30:56.000Z","pushType":"push","commitsCount":6,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[WebNN EP] Remove legacy MLOperandDescriptor.type (#20783)\n\nLatest Chrome has supported MLOperandDescriptor.dataType, remove legacy\r\nMLOperandDescriptor.type.","shortMessageHtmlLink":"[WebNN EP] Remove legacy MLOperandDescriptor.type (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2312325541\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20783\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20783/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20783\">microsoft#20783</a>)"}},{"before":"f1fef19b6e248023ec23b85bea8644ab2b95e831","after":"362a62390545234d28661307c7a203f2546280ae","ref":"refs/heads/main","pushedAt":"2024-05-29T00:20:23.000Z","pushType":"push","commitsCount":28,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"fix a build error with cuda 12.5 (#20770)\n\n### Motivation and Context\r\nhttps://github.com/microsoft/onnxruntime/issues/20765","shortMessageHtmlLink":"fix a build error with cuda 12.5 (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2311172047\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20770\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20770/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20770\">microsoft#20770</a>)"}},{"before":"478d3e0c62850952a7f5c1d54446f24139de4c8c","after":"f1fef19b6e248023ec23b85bea8644ab2b95e831","ref":"refs/heads/main","pushedAt":"2024-05-22T16:43:12.000Z","pushType":"push","commitsCount":62,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[js/webgpu] Support shared memory for transpose 2d (#19267)\n\nFor 1024x1024, without shared memoey, 18.7ms. With shared memory 13.2ms.","shortMessageHtmlLink":"[js/webgpu] Support shared memory for transpose 2d (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2099696039\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/19267\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/19267/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/19267\">microsoft#19267</a>)"}},{"before":"addcc4c4b296148e617435ac7ac917d278fa7a79","after":"478d3e0c62850952a7f5c1d54446f24139de4c8c","ref":"refs/heads/main","pushedAt":"2024-05-07T15:58:21.000Z","pushType":"push","commitsCount":4,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Add simplified layernorm fusion for Gemma (#20572)\n\nGemma has a `Mul` node right after the `Gather` and before the first\r\nlayer norm.","shortMessageHtmlLink":"Add simplified layernorm fusion for Gemma (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2280430668\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20572\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20572/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20572\">microsoft#20572</a>)"}},{"before":"589aeb7036eca41a01c9b76c93b21686fbdd5db9","after":"addcc4c4b296148e617435ac7ac917d278fa7a79","ref":"refs/heads/main","pushedAt":"2024-05-06T19:30:43.000Z","pushType":"push","commitsCount":10,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Fix missing node during mem efficient topo sort (#20497)\n\n### Fix missing node during mem efficient topo sort\r\n\r\nSome nodes are not cusumed by the backward path, they are also not\r\ngenerating graph outputs. We missed those nodes, so this PR fix that and\r\nadd related tests.\r\n\r\nA side note: we should remove those nodes that are not used for\r\ncomputing any graph outputs in a graph transformer. (TODO)\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"Fix missing node during mem efficient topo sort (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2267532514\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20497\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20497/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20497\">microsoft#20497</a>)"}},{"before":"218b6b0a734f6233845313d2c61531b7558e0b5f","after":"589aeb7036eca41a01c9b76c93b21686fbdd5db9","ref":"refs/heads/main","pushedAt":"2024-05-02T18:28:57.000Z","pushType":"push","commitsCount":45,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[DML EP] Register DFT-20 (#20341)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\nThis PR registers DFT-20 to the DML EP.\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"[DML EP] Register DFT-20 (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2246984250\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20341\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20341/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20341\">microsoft#20341</a>)"}},{"before":"cc3faba6161f8258451473a0dcc293bf0a18ad86","after":"218b6b0a734f6233845313d2c61531b7558e0b5f","ref":"refs/heads/main","pushedAt":"2024-04-25T00:10:53.000Z","pushType":"push","commitsCount":113,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Fix missing argument when calling _get_quantize_input_nodes (#20245)\n\n### Description\r\nThe current code is calling one method with a missing argument.\r\n\r\n\r\n\r\n### Motivation and Context\r\nIt breaks Olive's unittests.\r\n\r\n---------\r\n\r\nCo-authored-by: Xavier Dupré <xavier.dupre@gmail.com>","shortMessageHtmlLink":"Fix missing argument when calling _get_quantize_input_nodes (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2233668188\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20245\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20245/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20245\">microsof…</a>"}},{"before":"a6611409cc58efe1113331b2ba3188f2fd52697c","after":"cc3faba6161f8258451473a0dcc293bf0a18ad86","ref":"refs/heads/main","pushedAt":"2024-04-08T02:46:12.000Z","pushType":"push","commitsCount":4,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Support seq_len > 64K in rotary embedding cuda kernel (#20204)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"Support seq_len &gt; 64K in rotary embedding cuda kernel (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2226311786\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20204\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20204/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20204\">microsoft#20204</a>)"}},{"before":"3076b569472d0cbdae5e3657e3c267a63830b2b3","after":"a6611409cc58efe1113331b2ba3188f2fd52697c","ref":"refs/heads/main","pushedAt":"2024-04-05T18:17:11.000Z","pushType":"push","commitsCount":74,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Fix HalideIR title in third party notices reference (#20190)","shortMessageHtmlLink":"Fix HalideIR title in third party notices reference (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2222768282\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20190\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20190/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20190\">microsoft#20190</a>)"}},{"before":"06fe4f31131a6873a295ba47ed60f4cb16584296","after":"3076b569472d0cbdae5e3657e3c267a63830b2b3","ref":"refs/heads/main","pushedAt":"2024-03-23T00:08:57.000Z","pushType":"push","commitsCount":10,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Make MS Debug engine SymInitialize() called as needed. (#20036)\n\n### Description\r\n<!-- Describe your changes. -->\r\nInitialize Symbol engine as needed with no duplicate calls.\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->\r\n  Currently absel library may call SymInitialize more than once\r\n  when shared libraries are involved. However, this can only be\r\n  called only once per process. Our debug_alloc also may call it\r\n  when enabled. This change enables intialization to proceed\r\n  only when needed with no duplicate effort.","shortMessageHtmlLink":"Make MS Debug engine SymInitialize() called as needed. (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2203342375\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20036\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20036/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20036\">microsoft#20036</a>)"}},{"before":"6ff31e06d5757779b9c8d53e9d02a3b62b3e3438","after":"06fe4f31131a6873a295ba47ed60f4cb16584296","ref":"refs/heads/main","pushedAt":"2024-03-21T15:58:11.000Z","pushType":"push","commitsCount":10,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Increase MNIST test tolerance (#20000)\n\n### Description\r\n\r\nFound multiple occurrence of failures:\r\n\r\nhttps://dev.azure.com/onnxruntime/onnxruntime/_build/results?buildId=1321061&view=logs&j=6df8fe70-7b8f-505a-8ef0-8bf93da2bac7&t=56a04c0b-9e7f-5c69-cb7b-c2a7b1a7392a&l=17537\r\n\r\nhttps://dev.azure.com/onnxruntime/onnxruntime/_build/results?buildId=1329701&view=logs&j=6df8fe70-7b8f-505a-8ef0-8bf93da2bac7&t=4f6ef737-111d-50d1-a46b-5f86d9a970bc&s=3618b4c0-1011-591a-85b8-671e72e2cff1\r\n\r\n1: [ RUN      ] ModelTests/ModelTest.Run/\r\ncuda__models_zoo_opset7_MNIST_model\r\n1: D:\\a\\_work\\1\\s\\onnxruntime\\test\\providers\\cpu\\model_tests.cc(358):\r\nerror: Expected equality of these values:\r\n1:   COMPARE_RESULT::SUCCESS\r\n1:     Which is: 4-byte object <00-00 00-00>\r\n1:   ret.first\r\n1:     Which is: 4-byte object <01-00 00-00>\r\n1: expected -2.33638 (c0158735), got -2.30239 (c0135a47), diff:\r\n0.0339923, tol=0.0243638 idx=9","shortMessageHtmlLink":"Increase MNIST test tolerance (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2198439794\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/20000\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/20000/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/20000\">microsoft#20000</a>)"}},{"before":"9acaf534a62050705d9b892a57ef0e8409fa62ec","after":"6ff31e06d5757779b9c8d53e9d02a3b62b3e3438","ref":"refs/heads/main","pushedAt":"2024-03-20T06:28:13.000Z","pushType":"push","commitsCount":109,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"[MoE] Add TP and Mixtral MoE (#19945)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\n1.Support Tensor Parallelism in ShardedMoE.\r\n2.Make necessary code changes to support Mixtral MoE.\r\n3.Fix a bug related to using IOBinding in test script.\r\n4.Fix the input size limitation\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"[MoE] Add TP and Mixtral MoE (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2189634376\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/19945\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/19945/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/19945\">microsoft#19945</a>)"}},{"before":"1c468a03b90aa8122d49b3148152a67b0519d36e","after":"9acaf534a62050705d9b892a57ef0e8409fa62ec","ref":"refs/heads/main","pushedAt":"2024-03-04T17:03:51.000Z","pushType":"push","commitsCount":29,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Benchmark - Updating llama-2 requirement files (#19716)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"Benchmark - Updating llama-2 requirement files (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2160683004\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/19716\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/19716/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/19716\">microsoft#19716</a>)"}},{"before":"9cb97ee507b9b45d4a896f663590083e7e7568ac","after":"1c468a03b90aa8122d49b3148152a67b0519d36e","ref":"refs/heads/main","pushedAt":"2024-02-27T17:32:23.000Z","pushType":"push","commitsCount":92,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Improve Nuget-CUDA-Packaging-Pipeline (#19668)\n\n### Description\r\n<!-- Describe your changes. -->\r\n* Publish the artifacts as late as possible\r\n* once published the artifacts are immutable, and any retry will fail if\r\nthey exist\r\n  * if any step fails after publishing the stage cannot be retried\r\n* use powershell to cleanup\r\n  * DeleteFiles is taking >30 mins and causing the stage to timeout\r\n  * powershell took < 1s\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->\r\nMake pipeline more robust","shortMessageHtmlLink":"Improve Nuget-CUDA-Packaging-Pipeline (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2156045313\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/19668\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/19668/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/19668\">microsoft#19668</a>)"}},{"before":"28ad3ff799e163bf5c854db03e87086a61e14256","after":"9cb97ee507b9b45d4a896f663590083e7e7568ac","ref":"refs/heads/main","pushedAt":"2024-02-12T18:46:39.000Z","pushType":"push","commitsCount":677,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Disable CPU EP's allocator's arena when address sanitizer is enabled (#19485)\n\n### Description\r\nDisable CPU EP's allocator's arena when address sanitizer is enabled,\r\nbecause it masks problems. For example, the code in\r\nonnxruntime/test/quantization/quantization_test.cc has a memory leak\r\nproblem: it allocated a buffer but didn't free it, but most memory leak\r\ncheck tool cannot detect that because the buffer was from an arena and\r\nthe arena was finally freed.\r\n\r\n### Motivation and Context\r\nProvider better memory leak check coverage.","shortMessageHtmlLink":"Disable CPU EP's allocator's arena when address sanitizer is enabled (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2127775336\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/19485\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/19485/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/19485\">m…</a>"}},{"before":"0f72739b6db129373d221483d61d6637ec11fb28","after":"28ad3ff799e163bf5c854db03e87086a61e14256","ref":"refs/heads/main","pushedAt":"2023-10-27T18:04:28.000Z","pushType":"push","commitsCount":18,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Fix stale bot issue (#18064)\n\n### Description\r\nPreviously used GitHub stale app is now deprecated, so I deleted that\r\nfile and added a new GitHub Actions workflow to automatically apply the\r\nstale label to inactive issues.\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->","shortMessageHtmlLink":"Fix stale bot issue (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1957916412\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/18064\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/18064/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/18064\">microsoft#18064</a>)"}},{"before":"ae74a517b62baa6d973e46b5b51ac9a640512c46","after":"0f72739b6db129373d221483d61d6637ec11fb28","ref":"refs/heads/main","pushedAt":"2023-10-26T15:30:57.000Z","pushType":"push","commitsCount":272,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Disable ccache for WinML build (#18104)\n\n### Description\r\nIt seems would resolve the timeout issue. \r\n\r\n\r\n### Motivation and Context","shortMessageHtmlLink":"Disable ccache for WinML build (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1962350963\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/18104\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/18104/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/18104\">microsoft#18104</a>)"}},{"before":"2da08c477aedc62fd9b700bd005594e0dd130f49","after":"ae74a517b62baa6d973e46b5b51ac9a640512c46","ref":"refs/heads/main","pushedAt":"2023-09-08T18:44:21.000Z","pushType":"push","commitsCount":51,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Run Nuget_Test_Linux_GPU in container (#17452)\n\n### Description\r\n<!-- Describe your changes. -->\r\n\r\n\r\n\r\n### Motivation and Context\r\n<!-- - Why is this change required? What problem does it solve?\r\n- If it fixes an open issue, please link to the issue here. -->\r\n\r\n### Verification\r\n\r\nhttps://dev.azure.com/aiinfra/Lotus/_build/results?buildId=351542&view=results","shortMessageHtmlLink":"Run Nuget_Test_Linux_GPU in container (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1886204376\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/17452\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/17452/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/17452\">microsoft#17452</a>)"}},{"before":"71da0824f3644e378cb2a70ce63f6e4e24044804","after":"2da08c477aedc62fd9b700bd005594e0dd130f49","ref":"refs/heads/main","pushedAt":"2023-08-30T18:07:53.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Add website publish placeholder (#17318)","shortMessageHtmlLink":"Add website publish placeholder (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1870334933\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/17318\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/17318/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/17318\">microsoft#17318</a>)"}},{"before":"8224891236ae612a3e6a59ea3420b944f54fae4f","after":"71da0824f3644e378cb2a70ce63f6e4e24044804","ref":"refs/heads/main","pushedAt":"2023-08-30T16:28:44.000Z","pushType":"push","commitsCount":4,"pusher":{"login":"titaiwangms","name":"Ti-Tai Wang","path":"/titaiwangms","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18010845?s=80&v=4"},"commit":{"message":"Upgrade binskim and fix an error in nuget packaging pipeline (#17340)\n\n### Description\r\nUpgrade binskim and fix an error in nuget packaging pipeline.","shortMessageHtmlLink":"Upgrade binskim and fix an error in nuget packaging pipeline (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1872673899\" data-permission-text=\"Title is private\" data-url=\"https://github.com/microsoft/onnxruntime/issues/17340\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/microsoft/onnxruntime/pull/17340/hovercard\" href=\"https://github.com/microsoft/onnxruntime/pull/17340\">microso…</a>"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEpMcK-QA","startCursor":null,"endCursor":null}},"title":"Activity · titaiwangms/onnxruntime"}