{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":283871761,"defaultBranch":"main","name":"torch-mlir","ownerLogin":"llvm","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2020-07-30T20:35:59.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/17149993?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1715312425.0","currentOid":""},"activityList":{"items":[{"before":"afca88a0581c7815ce77485eedafbbf506aefb87","after":"fc100a117ddc3291559b60bcf0bb48cb66fe159b","ref":"refs/heads/main","pushedAt":"2024-05-31T07:36:48.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"renxida","name":"Xida Ren (Cedar)","path":"/renxida","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/10362952?s=80&v=4"},"commit":{"message":"[MLIR][ONNX] Add OnnxToTorch support for Scatter Op (#3400)\n\nThis PR adds OnnxToTorch support for Scatter op","shortMessageHtmlLink":"[MLIR][ONNX] Add OnnxToTorch support for Scatter Op (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2324363793\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3400\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3400/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3400\">#3400</a>)"}},{"before":"4e05e2cd1e9cc07a736fea61a463278e6f6431f9","after":"afca88a0581c7815ce77485eedafbbf506aefb87","ref":"refs/heads/main","pushedAt":"2024-05-31T06:45:14.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"rsuderman","name":"Rob Suderman","path":"/rsuderman","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/5508949?s=80&v=4"},"commit":{"message":"[NFC] Change to *cast instead of .*cast variants (#3405)\n\nMember casts have been deprecated. Changing over a bunch of the member\r\ncast calls to the global templated variants to remove deprecation\r\nwarnings.","shortMessageHtmlLink":"[NFC] Change to *cast instead of .*cast variants (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2326994114\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3405\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3405/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3405\">#3405</a>)"}},{"before":"074098d20cd1f62ddfb8379bfc6f42530d6976df","after":"4e05e2cd1e9cc07a736fea61a463278e6f6431f9","ref":"refs/heads/main","pushedAt":"2024-05-31T01:56:48.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"[Torch] support recompose of aten.split.with_sizes and aten.tensor_sp… (#3401)\n\n…lit.sections\r\n\r\n* support recompose to aten.split.with_sizes and\r\naten.tensor_split.sections\r\n* fix recompose of aten.chunk","shortMessageHtmlLink":"[Torch] support recompose of aten.split.with_sizes and aten.tensor_sp… ("}},{"before":"d7b8f00d017253b98d5c41aa120938678f8ec672","after":"074098d20cd1f62ddfb8379bfc6f42530d6976df","ref":"refs/heads/main","pushedAt":"2024-05-31T00:34:38.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"renxida","name":"Xida Ren (Cedar)","path":"/renxida","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/10362952?s=80&v=4"},"commit":{"message":"Modifies onnx resize lowering to fix numerical issues (#3381)\n\nUpdates:\r\n\r\n- some unsupported modes are now going to report a match failure for\r\nunsupported coordinate transformation modes.\r\n- fixes a bug that was introduced in the last patch for resize (my\r\nbad...)\r\n- uses actual x and y coordinates for computing weights in bilinear\r\ninterpolation (rather than eps modified values)\r\n- slightly simplifies the bilinear interpolation payload for readability\r\nand performance\r\n- passes coordinate transformation mode information from an onnx.Resize\r\nop to the mode string for the aten._interpolate op. This allows us to\r\nperform custom logic in the torch->linalg lowering to support\r\nonnx.Resize options without losing the default behaviors of the\r\ninterpolate op.","shortMessageHtmlLink":"Modifies onnx resize lowering to fix numerical issues (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2311500817\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3381\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3381/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3381\">#3381</a>)"}},{"before":"e4be197efd85916cd378ef8e7f21ca3de13b5903","after":"d7b8f00d017253b98d5c41aa120938678f8ec672","ref":"refs/heads/main","pushedAt":"2024-05-30T17:35:26.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"vivekkhandelwal1","name":"Vivek Khandelwal","path":"/vivekkhandelwal1","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/68822896?s=80&v=4"},"commit":{"message":"[ONNX] Add OnnxToTorch Lowering for LpNormalization op (#3397)\n\nSigned-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>","shortMessageHtmlLink":"[ONNX] Add OnnxToTorch Lowering for LpNormalization op (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2323821636\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3397\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3397/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3397\">#3397</a>)"}},{"before":"1f544c37d0fb9f9657e4f80e9c30ccad8e3e0dc2","after":"e4be197efd85916cd378ef8e7f21ca3de13b5903","ref":"refs/heads/main","pushedAt":"2024-05-30T06:31:18.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[FxImporter] Fix transpose rank zero (#3382)","shortMessageHtmlLink":"[FxImporter] Fix transpose rank zero (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2312129201\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3382\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3382/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3382\">#3382</a>)"}},{"before":"23d2d66a59cc85e13ed1c9a6c4875b65c1b2db42","after":"1f544c37d0fb9f9657e4f80e9c30ccad8e3e0dc2","ref":"refs/heads/main","pushedAt":"2024-05-30T06:30:36.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[NFC] Remove unused header files (#3386)","shortMessageHtmlLink":"[NFC] Remove unused header files (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2313243154\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3386\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3386/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3386\">#3386</a>)"}},{"before":"a5d3b546f86f9547049b5ee4562bb4698de50b38","after":"23d2d66a59cc85e13ed1c9a6c4875b65c1b2db42","ref":"refs/heads/main","pushedAt":"2024-05-29T23:56:23.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"renxida","name":"Xida Ren (Cedar)","path":"/renxida","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/10362952?s=80&v=4"},"commit":{"message":"Fix error when attempting to read elided onnx constants (#3398)\n\nCo-authored-by: zjgarvey <zjgarvey@gmail.com>","shortMessageHtmlLink":"Fix error when attempting to read elided onnx constants (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2323921026\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3398\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3398/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3398\">#3398</a>)"}},{"before":"e0a5adb1db38b0072c44b87570bc530eb3b324ad","after":"a5d3b546f86f9547049b5ee4562bb4698de50b38","ref":"refs/heads/main","pushedAt":"2024-05-29T06:46:21.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[FxImporter] Fix embedding bag (#3387)","shortMessageHtmlLink":"[FxImporter] Fix embedding bag (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2315201273\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3387\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3387/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3387\">#3387</a>)"}},{"before":"05929f9171d2c21316fc41e4a67508f6bbcea4a4","after":"e0a5adb1db38b0072c44b87570bc530eb3b324ad","ref":"refs/heads/main","pushedAt":"2024-05-27T07:49:51.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"[Torch] fix aten.linear's decomposition (#3391)\n\n* support aten.linear with more rank.","shortMessageHtmlLink":"[Torch] fix aten.linear's decomposition (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2318349458\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3391\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3391/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3391\">#3391</a>)"}},{"before":"28aeb047c1ae50b8d6911c35fc0d323e7dd862fb","after":"05929f9171d2c21316fc41e4a67508f6bbcea4a4","ref":"refs/heads/main","pushedAt":"2024-05-27T00:01:07.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"enhance verbose option in e2e_testing (#3390)\n\nso that `python3 e2e_testing/main.py -v` would print intermediate IR.","shortMessageHtmlLink":"enhance verbose option in e2e_testing (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2317818249\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3390\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3390/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3390\">#3390</a>)"}},{"before":"27169dcda993320448549f2d95ebc3c1ea38d111","after":"28aeb047c1ae50b8d6911c35fc0d323e7dd862fb","ref":"refs/heads/main","pushedAt":"2024-05-26T04:34:56.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"[Stablehlo] fix crashing on AtenEmbeddingBagSumExample_basic (#3389)","shortMessageHtmlLink":"[Stablehlo] fix crashing on AtenEmbeddingBagSumExample_basic (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2317159227\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3389\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3389/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3389\">#3389</a>)"}},{"before":"5bb1a65ec93dbef78fd2703a050bea1ab6fa6994","after":"27169dcda993320448549f2d95ebc3c1ea38d111","ref":"refs/heads/main","pushedAt":"2024-05-23T16:01:47.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ScottTodd","name":"Scott Todd","path":"/ScottTodd","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/4010439?s=80&v=4"},"commit":{"message":"Replace some depreciated uses of cast (#3343)\n\nContributing towards #3299","shortMessageHtmlLink":"Replace some depreciated uses of cast (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2295932337\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3343\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3343/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3343\">#3343</a>)"}},{"before":"43f961eca449ad32c892455162ace897548170d8","after":"5bb1a65ec93dbef78fd2703a050bea1ab6fa6994","ref":"refs/heads/main","pushedAt":"2024-05-23T12:40:20.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"[Stablehlo] refactor reduction lowering and support aten.amin (#3383)\n\n* implement detailed lowering template pattern\r\n`ConvertAtenReduceAllDimsOp` and `ConvertAtenReduceKeepDimOp`\r\n* support `aten.amin`'s lowering.","shortMessageHtmlLink":"[Stablehlo] refactor reduction lowering and support aten.amin (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2312316973\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3383\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3383/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3383\">#3383</a>)"}},{"before":"d924d0047fc608fbb1c1751a11a655601c4a2131","after":"43f961eca449ad32c892455162ace897548170d8","ref":"refs/heads/main","pushedAt":"2024-05-23T03:29:28.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"pashu123","name":"Prashant Kumar","path":"/pashu123","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/16246821?s=80&v=4"},"commit":{"message":"[MLIR] Fix 64-bit product during aten.view lowering (#3378)\n\nstd::accumulate needs 64-bit init value to perform 64-bit arithmetic on\r\na list of integers.\r\n\r\nSigned-off-by: Gaurav Shukla <gaurav.shukla@amd.com>","shortMessageHtmlLink":"[MLIR] Fix 64-bit product during aten.view lowering (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2310992000\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3378\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3378/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3378\">#3378</a>)"}},{"before":"2e194e13d69cc1a839509761fbc193952ca8f9bc","after":"d924d0047fc608fbb1c1751a11a655601c4a2131","ref":"refs/heads/main","pushedAt":"2024-05-23T01:55:33.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[FxImporter] Fix primitive type in return (#3379)","shortMessageHtmlLink":"[FxImporter] Fix primitive type in return (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2311247695\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3379\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3379/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3379\">#3379</a>)"}},{"before":"f4bfe3f948a11abc308c9fe46c571ee67406e109","after":"2e194e13d69cc1a839509761fbc193952ca8f9bc","ref":"refs/heads/main","pushedAt":"2024-05-22T17:19:09.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"kuhar","name":"Jakub Kuderski","path":"/kuhar","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/4612584?s=80&v=4"},"commit":{"message":"[Torch] Fix bugs for `Torch::AtenOneHotOp` (#3350)\n\nThis PR fixes the bugs for `Torch::AtenOneHotOp` by:\r\n\r\n1) Using `Torch::kUnknownSize` as the default value for `numClasses` in\r\n   the pattern matching stage in `DecomposeAtenOneHotOp`\r\n2) Adding `AtenIntScalarOp` to the patterns in `TorchToArith`\r\n3) Handling both `int` and `float` types for `off` and `on` values in\r\n`TorchOnnxToTorch` conversion\r\n\r\nIt also includes:\r\n\r\n1) A new test in `TorchToArith/basic.mlir`, for `torch.aten.Int.Scalar`,\r\nand\r\n2) A new test in `decompose-complex-ops.mlir`, for `torch.aten.one_hot`\r\n\r\n**Dependencies**\r\n\r\nThis PR is dependent on #3334.","shortMessageHtmlLink":"[Torch] Fix bugs for <code>Torch::AtenOneHotOp</code> (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2298844214\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3350\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3350/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3350\">#3350</a>)"}},{"before":"4d7cdba4bf29b3665094b843550917430a845a10","after":"f4bfe3f948a11abc308c9fe46c571ee67406e109","ref":"refs/heads/main","pushedAt":"2024-05-22T15:28:45.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"Bump llvm and stablehlo (#3377)\n\n* bump llvm to 1e5f29af81a5f6fda308074f6345b9fba4faa71c\r\n* bump stablehlo to c44d9af8d4879adccf1054cb61a53377ae5898cb","shortMessageHtmlLink":"Bump llvm and stablehlo (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2310627325\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3377\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3377/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3377\">#3377</a>)"}},{"before":"972d47b58617050fd00a7dc420f18bfea81ab364","after":"4d7cdba4bf29b3665094b843550917430a845a10","ref":"refs/heads/main","pushedAt":"2024-05-22T15:16:57.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qingyunqu","name":"Yuanqiang Liu","path":"/qingyunqu","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/29956693?s=80&v=4"},"commit":{"message":"[Torch] eliminate \"getWithLeastStaticInformation\" in DecomposeAtenTriuOp (#3330)\n\nI am trying to eliminate 'getWithLeastStaticInformation' in\r\nDecomposeAtenTriuOp. Could you provide me with some suggestions?\r\n@qingyunqu @zjgarvey \r\nSee issue https://github.com/llvm/torch-mlir/issues/3312","shortMessageHtmlLink":"[Torch] eliminate \"getWithLeastStaticInformation\" in DecomposeAtenTri…"}},{"before":"52be4bdc188e1f6751b56483e19ee87a3b9a140e","after":"972d47b58617050fd00a7dc420f18bfea81ab364","ref":"refs/heads/main","pushedAt":"2024-05-22T14:59:01.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[FxImporter] Fix constant bool tensor (#3375)","shortMessageHtmlLink":"[FxImporter] Fix constant bool tensor (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2309493732\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3375\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3375/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3375\">#3375</a>)"}},{"before":"6e485574e55cadc441470457e49470f5e6ac54d0","after":"52be4bdc188e1f6751b56483e19ee87a3b9a140e","ref":"refs/heads/main","pushedAt":"2024-05-22T12:32:00.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"kuhar","name":"Jakub Kuderski","path":"/kuhar","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/4612584?s=80&v=4"},"commit":{"message":"[ONNX] Fix bugs for the `onnx.OneHot` operator (#3334)\n\nThis commit fixes the bugs for the `onnx.OneHot` operator by:\r\n\r\n1) Converting negative indices to non-negative indices\r\n2) Handling both `int` and `float` types for `off` and `on` values\r\n3) Using the correct result type\r\n\r\nIt also includes a new unit test.","shortMessageHtmlLink":"[ONNX] Fix bugs for the <code>onnx.OneHot</code> operator (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2293734214\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3334\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3334/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3334\">#3334</a>)"}},{"before":"560ca24771fecdc516cbfb580c9d2459c6365c6f","after":"6e485574e55cadc441470457e49470f5e6ac54d0","ref":"refs/heads/main","pushedAt":"2024-05-22T12:23:18.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"sjain-stanford","name":"Sambhav Jain","path":"/sjain-stanford","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/19234106?s=80&v=4"},"commit":{"message":"[Pipeline] Use dedicated simplification pipeline for TorchDynamo frontend (#3376)\n\nDiscord Thread:\r\nhttps://discord.com/channels/636084430946959380/1238330633328005243\r\n\r\n## Context: \r\n\r\n[This](https://github.com/llvm/torch-mlir/blob/main/python/torch_mlir/fx.py#L61)\r\nwas updated to support e2e tests for the TorchDynamo frontend in\r\nTorch-MLIR, where we run FX decompositions and import the FX IR to\r\ngenerate Torch dialect, followed by\r\n`torch-function-to-torch-backend-pipeline`, skipping only the shape/type\r\nrefinement for now. However, we should be able to skip many of the torch\r\nsimplification passes, as depicted in the [frontend\r\nroadmap](https://github.com/llvm/torch-mlir/blob/main/docs/images/roadmap_frontend.png).\r\n\r\nBased on IREE's TorchDynamo\r\n[pipeline](https://github.com/iree-org/iree/blob/main/compiler/plugins/input/Torch/InputConversion/Passes.cpp#L29),\r\nthe only two passes we seem to require are: `ReduceOpVariantsPass` and\r\n`DecomposeComplexOpsPass`. This is inline with our findings as well\r\nbased on initial exploration.\r\n\r\nThis PR creates a dedicated frontend simplification pipeline for\r\nTorchDynamo / FX Importer which calls only `ReduceOpVariantsPass` and\r\n`DecomposeComplexOpsPass`. We rely on the e2e fx_importer tests to\r\nensure we're not regressing by removing many of the passes that were\r\nhistorically needed for TorchScript.\r\n\r\nOne notable change here is that we do not call the\r\n`LowerToBackendContractPass` anymore, which used to call\r\n`TorchSimplificationPipeline` iteratively until VerifyBackendContract\r\nwas clean. Some of this was required for the shape/type refinement to\r\nconverge, which seems a non-issue for Dynamo frontend. Do we anticipate\r\nthis (the iterative invocation of TorchSimplificationPipeline followed\r\nby VerifyBackendContract) to be worth retaining in the Dynamo frontend\r\npipeline? If so, I can make those changes, PLMK.","shortMessageHtmlLink":"[Pipeline] Use dedicated simplification pipeline for TorchDynamo fron…"}},{"before":"fcf48872b30278d4a84f70d646d82a13e7096edb","after":"560ca24771fecdc516cbfb580c9d2459c6365c6f","ref":"refs/heads/main","pushedAt":"2024-05-22T00:12:55.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"aartbik","name":"Aart Bik","path":"/aartbik","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/39774503?s=80&v=4"},"commit":{"message":"[torch-mlir][sparse] replace xavier with ones initialization (#3374)\n\nensures stability of results between different set ups","shortMessageHtmlLink":"[torch-mlir][sparse] replace xavier with ones initialization (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2309239706\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3374\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3374/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3374\">#3374</a>)"}},{"before":"c2c1c2cfa40e89d3acb79c7ccf720ad539e16c56","after":"fcf48872b30278d4a84f70d646d82a13e7096edb","ref":"refs/heads/main","pushedAt":"2024-05-21T19:10:26.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"rsuderman","name":"Rob Suderman","path":"/rsuderman","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/5508949?s=80&v=4"},"commit":{"message":"[ONNX] Implement Softsign op (#3373)","shortMessageHtmlLink":"[ONNX] Implement Softsign op (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2308778741\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3373\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3373/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3373\">#3373</a>)"}},{"before":"b870729efe5929b1ee6ff1c7b27d4d1857cdacc7","after":"c2c1c2cfa40e89d3acb79c7ccf720ad539e16c56","ref":"refs/heads/main","pushedAt":"2024-05-21T16:20:54.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"penguin-wwy","name":"penguin_wwy","path":"/penguin-wwy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/15888598?s=80&v=4"},"commit":{"message":"[FxImporter] Fix failed e2e case (#3365)","shortMessageHtmlLink":"[FxImporter] Fix failed e2e case (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2304651887\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3365\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3365/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3365\">#3365</a>)"}},{"before":"c0e7d2667dc2afae19c89946f1b5c9855d726b94","after":"b870729efe5929b1ee6ff1c7b27d4d1857cdacc7","ref":"refs/heads/main","pushedAt":"2024-05-21T15:35:33.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"vivekkhandelwal1","name":"Vivek Khandelwal","path":"/vivekkhandelwal1","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/68822896?s=80&v=4"},"commit":{"message":"[torch] Fix `onnx.MaxPool` lowering (#3133)\n\nThis commit fixes the onnx.MaxPool op lowering which was lacking the\r\nindices result support.\r\n\r\nSigned-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>","shortMessageHtmlLink":"[torch] Fix <code>onnx.MaxPool</code> lowering (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2235974210\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3133\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3133/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3133\">#3133</a>)"}},{"before":"297c270980f7affb3cb075c902724dfbf2eaf3cf","after":"c0e7d2667dc2afae19c89946f1b5c9855d726b94","ref":"refs/heads/main","pushedAt":"2024-05-21T02:52:17.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"aartbik","name":"Aart Bik","path":"/aartbik","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/39774503?s=80&v=4"},"commit":{"message":"[torch-mlir][sparse] inference mode for sparse GCN test (#3369)","shortMessageHtmlLink":"[torch-mlir][sparse] inference mode for sparse GCN test (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2306986564\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3369\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3369/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3369\">#3369</a>)"}},{"before":"99511cef82997fd9faf6c5d2be4b932a37ec0f96","after":"297c270980f7affb3cb075c902724dfbf2eaf3cf","ref":"refs/heads/main","pushedAt":"2024-05-20T20:35:27.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"rsuderman","name":"Rob Suderman","path":"/rsuderman","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/5508949?s=80&v=4"},"commit":{"message":"onnx.Resize and aten._interpolate : allow n spatial dims.  (#3368)\n\nThe old lowering only had logic for 2d (i.e. images). this patch allows\r\ninterpolation for n spatial dims, which is required for some 3d vision\r\nmodels such as\r\n\r\n- onnx/models/pytorch-3dunet_vaiq_int8\r\n\r\nwhich successfully compiles and runs with this patch.","shortMessageHtmlLink":"onnx.Resize and aten._interpolate : allow n spatial dims. (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2306558389\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3368\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3368/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3368\">#3368</a>)"}},{"before":"cc28d566ff02904bde4855f3e2c8a124ecb6f4d6","after":"99511cef82997fd9faf6c5d2be4b932a37ec0f96","ref":"refs/heads/main","pushedAt":"2024-05-20T15:26:24.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"vivekkhandelwal1","name":"Vivek Khandelwal","path":"/vivekkhandelwal1","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/68822896?s=80&v=4"},"commit":{"message":"Implement `onnx.Hardmax` lowering (#3342)\n\nCo-authored-by: Ubuntu <xunli@wsno1.judsoscro3wupi0qm4bjlj5m3b.bx.internal.cloudapp.net>\r\nCo-authored-by: Hasekawa-Takumi <bewater.private476@passmail.net>","shortMessageHtmlLink":"Implement <code>onnx.Hardmax</code> lowering (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2295869568\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3342\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3342/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3342\">#3342</a>)"}},{"before":"8814d0ae64e2e276e5f23186dc364dc77505678f","after":"cc28d566ff02904bde4855f3e2c8a124ecb6f4d6","ref":"refs/heads/main","pushedAt":"2024-05-20T07:49:24.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"qiuxiafei","name":"Xiafei Qiu","path":"/qiuxiafei","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/261796?s=80&v=4"},"commit":{"message":"[Stablehlo] Support AtenTrilOp (#3359)\n\n1. lower aten.tril to stablehlo composed by iota, select and so forth\r\n2. add related e2e test cases","shortMessageHtmlLink":"[Stablehlo] Support AtenTrilOp (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2302099603\" data-permission-text=\"Title is private\" data-url=\"https://github.com/llvm/torch-mlir/issues/3359\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/llvm/torch-mlir/pull/3359/hovercard\" href=\"https://github.com/llvm/torch-mlir/pull/3359\">#3359</a>)"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEWMFEpgA","startCursor":null,"endCursor":null}},"title":"Activity · llvm/torch-mlir"}