{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":154739597,"defaultBranch":"main","name":"jax","ownerLogin":"google","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2018-10-25T21:25:02.000Z","ownerAvatar":"https://github.com/avatars/u/1342004?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1719724441.0","currentOid":""},"activityList":{"items":[{"before":"5fac179f2fb0006588943cbad1525371e601a996","after":"1949691daabe815f4b098253609dc4912b3d61d8","ref":"refs/heads/main","pushedAt":"2024-06-30T18:59:12.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Update XLA dependency to use revision\nhttp://github.com/openxla/xla/commit/6fd83ac4474a58b3afd5ec832612e59a441153f4.\n\nPiperOrigin-RevId: 648165885","shortMessageHtmlLink":"Update XLA dependency to use revision"}},{"before":"89c404e703a111b8984bffbe68cd2dfb4254d545","after":"5fac179f2fb0006588943cbad1525371e601a996","ref":"refs/heads/main","pushedAt":"2024-06-30T16:15:21.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Merge pull request #22134 from gnecula:pallas_doc\n\nPiperOrigin-RevId: 648147118","shortMessageHtmlLink":"Merge pull request <a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"2377474130\" data-permission-text=\"Title is private\" data-url=\"https://github.com/google/jax/issues/22134\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/google/jax/pull/22134/hovercard\" href=\"https://github.com/google/jax/pull/22134\">#22134</a> from gnecula:pallas_doc"}},{"before":"d72fa91c80725629d56bc11ef65a7d2246e0f3d2","after":"51354f6a1622ff4cf62a09af487bbc6560e79546","ref":"refs/heads/test_648066015","pushedAt":"2024-06-30T05:19:10.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"[pallas] Make pallas_test run on CPU and TPU also.\n\npallas_test was only running on GPU, but it is useful to run this test on all platform, in both interpret mode and the native mode. Added `skipTest` and `TODO` for the tests that fail, and in some cases configured numerical comparison tolerances.\n\nAll tests now have a \"Interpreter\" version, e.g., for `CallTest` we also define a `CallInterpreterTest` that runs the same tests but in interpreter\nmode. This was not done systematically before, and in some cases the\ninterpreter test was missing, or was empty.\n\nSome of the tests in pallas_test perhaps make sense only for GPU. I will\nsplit them out in a separate CL.\n\nPiperOrigin-RevId: 648066015","shortMessageHtmlLink":"[pallas] Make pallas_test run on CPU and TPU also."}},{"before":"07b0c045d71f13328aebb30b04e58d4cb1959409","after":"d72fa91c80725629d56bc11ef65a7d2246e0f3d2","ref":"refs/heads/test_648066015","pushedAt":"2024-06-30T05:18:10.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"[pallas] Make pallas_test run on CPU and TPU also.\n\npallas_test was only running on GPU, but it is useful to run this test on all platform, in both interpret mode and the native mode. Added `skipTest` and `TODO` for the tests that fail, and in some cases configured numerical comparison tolerances.\n\nAll tests now have a \"Interpreter\" version, e.g., for `CallTest` we also define a `CallInterpreterTest` that runs the same tests but in interpreter\nmode. This was not done systematically before, and in some cases the\ninterpreter test was missing, or was empty.\n\nSome of the tests in pallas_test perhaps make sense only for GPU. I will\nsplit them out in a separate CL.\n\nPiperOrigin-RevId: 648066015","shortMessageHtmlLink":"[pallas] Make pallas_test run on CPU and TPU also."}},{"before":null,"after":"07b0c045d71f13328aebb30b04e58d4cb1959409","ref":"refs/heads/test_648066015","pushedAt":"2024-06-30T05:14:01.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"[pallas] Make pallas_test run on CPU and TPU also.\n\npallas_test was only running on GPU, but it is useful to run this test on all platform, in both interpret mode and the native mode. Added `skipTest` and `TODO` for the tests that fail, and in some cases configured numerical comparison tolerances.\n\nAll tests now have a \"Interpreter\" version, e.g., for `CallTest` we also define a `CallInterpreterTest` that runs the same tests but in interpreter\nmode. This was not done systematically before, and in some cases the\ninterpreter test was missing, or was empty.\n\nSome of the tests in pallas_test perhaps make sense only for GPU. I will\nsplit them out in a separate CL.\n\nPiperOrigin-RevId: 648066015","shortMessageHtmlLink":"[pallas] Make pallas_test run on CPU and TPU also."}},{"before":"89c404e703a111b8984bffbe68cd2dfb4254d545","after":null,"ref":"refs/heads/test_648004327","pushedAt":"2024-06-29T21:36:49.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"}},{"before":"2f808e9da902063bcd85e5ef2e4f16b95ddc9255","after":"89c404e703a111b8984bffbe68cd2dfb4254d545","ref":"refs/heads/main","pushedAt":"2024-06-29T21:36:48.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Improve error message when a global jax.Array is closed over a jitted function in McJAX.\n\nPiperOrigin-RevId: 648010704","shortMessageHtmlLink":"Improve error message when a global jax.Array is closed over a jitted…"}},{"before":"130e97d843b5d991d102b359afc2ad078ccfdcff","after":"89c404e703a111b8984bffbe68cd2dfb4254d545","ref":"refs/heads/test_648004327","pushedAt":"2024-06-29T21:36:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Improve error message when a global jax.Array is closed over a jitted function in McJAX.\n\nPiperOrigin-RevId: 648010704","shortMessageHtmlLink":"Improve error message when a global jax.Array is closed over a jitted…"}},{"before":"73fc4d5620641c7502c64fd3aaaf4861510d9094","after":"130e97d843b5d991d102b359afc2ad078ccfdcff","ref":"refs/heads/test_648004327","pushedAt":"2024-06-29T21:23:33.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Improve error message when a global jax.Array is closed over a jitted function in McJAX.\n\nPiperOrigin-RevId: 648004327","shortMessageHtmlLink":"Improve error message when a global jax.Array is closed over a jitted…"}},{"before":"cb225b4a8648b0c50fa6178839d87e3d66d6105d","after":"73fc4d5620641c7502c64fd3aaaf4861510d9094","ref":"refs/heads/test_648004327","pushedAt":"2024-06-29T20:43:09.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Improve error message when a global jax.Array is closed over a jitted function in McJAX.\n\nPiperOrigin-RevId: 648004327","shortMessageHtmlLink":"Improve error message when a global jax.Array is closed over a jitted…"}},{"before":null,"after":"cb225b4a8648b0c50fa6178839d87e3d66d6105d","ref":"refs/heads/test_648004327","pushedAt":"2024-06-29T20:42:16.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Improve error message when a global jax.Array is closed over a jitted function in McJAX.\n\nPiperOrigin-RevId: 648004327","shortMessageHtmlLink":"Improve error message when a global jax.Array is closed over a jitted…"}},{"before":"2f808e9da902063bcd85e5ef2e4f16b95ddc9255","after":null,"ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T19:18:39.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"}},{"before":"2e83c324373052e1b58dcd3ecb58065f5d209e59","after":"2f808e9da902063bcd85e5ef2e4f16b95ddc9255","ref":"refs/heads/main","pushedAt":"2024-06-29T19:18:38.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647993991","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"a2c8ccd152f27f8a0478b87b938d40ef9ee27cf3","after":"2f808e9da902063bcd85e5ef2e4f16b95ddc9255","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T19:18:37.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647993991","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"071da567fb685708cc75528c2a6a1bcb06734275","after":"2e83c324373052e1b58dcd3ecb58065f5d209e59","ref":"refs/heads/main","pushedAt":"2024-06-29T18:57:56.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Update XLA dependency to use revision\nhttp://github.com/openxla/xla/commit/46b9412435ada4ae326284aa977f223ae6f40990.\n\nPiperOrigin-RevId: 647991056","shortMessageHtmlLink":"Update XLA dependency to use revision"}},{"before":"0c66f9bf189abda5da42a5533c3097e9b3a53c2f","after":"a2c8ccd152f27f8a0478b87b938d40ef9ee27cf3","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T18:55:37.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"17f957e3d37e092674aec561bc2b77772dc90981","after":"0c66f9bf189abda5da42a5533c3097e9b3a53c2f","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T15:40:03.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"1f8d5998425e1df6b13b73c2741780e2fedada8d","after":"17f957e3d37e092674aec561bc2b77772dc90981","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T08:23:55.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"cc20f93f906af28af6baf40f515521f26589a2b2","after":"1f8d5998425e1df6b13b73c2741780e2fedada8d","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T08:21:20.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error in custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error in custom call registration for some FFI functions"}},{"before":"42443b960e618deff0605a5b49ed7d29c9030453","after":"cc20f93f906af28af6baf40f515521f26589a2b2","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T06:48:02.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error regarding custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error regarding custom call registration for some FFI functions"}},{"before":"ae1537d9223a2716fb21e62d3928fd663493eaa3","after":"42443b960e618deff0605a5b49ed7d29c9030453","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T06:46:20.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error regarding custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error regarding custom call registration for some FFI functions"}},{"before":"6ace4e350949601628c8fb37f3d19e225c6e7fc2","after":"ae1537d9223a2716fb21e62d3928fd663493eaa3","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T06:29:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error regarding custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to rollback the changes in https://github.com/google/jax/pull/22178\nand disable the changes from https://github.com/google/jax/pull/20997.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error regarding custom call registration for some FFI functions"}},{"before":"709496cfc17447fce6e8f99f5213bafa3db7bd35","after":"6ace4e350949601628c8fb37f3d19e225c6e7fc2","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T06:19:37.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Fix error regarding custom call registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to use the old style registration calls from C++.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Fix error regarding custom call registration for some FFI functions"}},{"before":"7c8f81df8b11adce0dacb32f582949d16ecdb54f","after":"709496cfc17447fce6e8f99f5213bafa3db7bd35","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T05:59:02.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Remove the Python-side kernel registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to remove the Python-side registration, assuming that\nthe call targets will be registered by the library.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Remove the Python-side kernel registration for some FFI functions"}},{"before":"7d41ecbbf24e8cef532fbd10996c10e6ff5ad38d","after":"7c8f81df8b11adce0dacb32f582949d16ecdb54f","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T05:42:02.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Remove the Python-side kernel registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to remove the Python-side registration, assuming that\nthe call targets will be registered by the library.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Remove the Python-side kernel registration for some FFI functions"}},{"before":"619aca06a7c1993a73218167790a98fa98267ba2","after":"7d41ecbbf24e8cef532fbd10996c10e6ff5ad38d","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T05:37:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Remove the Python-side kernel registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to remove the Python-side registration, assuming that\nthe call targets will be registered by the library.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Remove the Python-side kernel registration for some FFI functions"}},{"before":"b9a5f971e4d0a4a88db6f4940be2638c82c0aa5c","after":"619aca06a7c1993a73218167790a98fa98267ba2","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T05:19:35.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Remove the Python-side kernel registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to remove the Python-side registration, assuming that\nthe call targets will be registered by the library.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Remove the Python-side kernel registration for some FFI functions"}},{"before":null,"after":"b9a5f971e4d0a4a88db6f4940be2638c82c0aa5c","ref":"refs/heads/test_647886177","pushedAt":"2024-06-29T05:18:34.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Remove the Python-side kernel registration for some FFI functions\n\nWe are getting the following errors:\n```\nDuplicate FFI handler registration for cu_threefry2x32_ffi on a platform CUDA\nDuplicate FFI handler registration for cu_lu_pivots_to_permutation on a platform CUDA\n```\n\nIt seems that with the ffi registration mechanism based on `XLA_FFI_REGISTER_HANDLER` it is not possible anymore to\nregister a call target twice.\n\nThe fix here is to remove the Python-side registration, assuming that\nthe call targets will be registered by the library.\n\nPiperOrigin-RevId: 647886177","shortMessageHtmlLink":"Remove the Python-side kernel registration for some FFI functions"}},{"before":"4e8d813d87a1df362f142ad694532e408631b18b","after":"691c96f4cf2b9aae3a18995818ed9fed21a44692","ref":"refs/heads/adx4","pushedAt":"2024-06-29T02:36:21.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"dougalm","name":"Dougal Maclaurin","path":"/dougalm","primaryAvatarUrl":"https://github.com/avatars/u/8833761?s=80&v=4"},"commit":{"message":"[adx4] Add array and tuple types","shortMessageHtmlLink":"[adx4] Add array and tuple types"}},{"before":"9e71c84955f85af4ad3f3e10b57e98e1dea6e6e7","after":"1ec2f087a00a2f09e7c8c15cf26c332fb0809a24","ref":"refs/heads/test_616865795","pushedAt":"2024-06-28T23:58:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"copybara-service[bot]","name":null,"path":"/apps/copybara-service","primaryAvatarUrl":"https://github.com/avatars/in/44061?s=80&v=4"},"commit":{"message":"Introduce hermetic CUDA in Google ML projects.\n\nInstead of having pre-installed NVIDIA CUDA and CUDNN libraries and setting environment variables pointing to the installation locations, Bazel should automatically download CUDA and CUDNN distributives in the cache and use them during build and test phases.\n\nPiperOrigin-RevId: 616865795","shortMessageHtmlLink":"Introduce hermetic CUDA in Google ML projects."}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEcyBJQgA","startCursor":null,"endCursor":null}},"title":"Activity · google/jax"}