OSSForks/tensorflow
1
0
Fork 0
mirror of https://github.com/zebrajr/tensorflow.git synced 2025-12-06 00:19:58 +01:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Commit Graph

  • ac51b90ed6 Automated Code Change master A. Unique TensorFlower 2025-11-01 03:44:27 -0700
  • 4f3f2c9444 [XLA:GPU] add NanCount thunk to thunk_buffer_debug_pass Ilya Tikhonovskiy 2025-11-01 03:05:12 -0700
  • 459ba30568 compat: Update forward compatibility horizon to 2025-11-01 A. Unique TensorFlower 2025-11-01 02:03:52 -0700
  • ea390babb6 Update GraphDef version to 2398. A. Unique TensorFlower 2025-11-01 02:03:46 -0700
  • ef28899305 Automated Code Change A. Unique TensorFlower 2025-10-31 22:30:17 -0700
  • aa4db17b00 Automated Code Change A. Unique TensorFlower 2025-10-31 20:31:50 -0700
  • 09f85795ba Automated Code Change A. Unique TensorFlower 2025-10-31 20:24:02 -0700
  • 4618f903c4 Reverts bec8916f32 A. Unique TensorFlower 2025-10-31 20:10:47 -0700
  • 752a654e9e [jax:ffi] Declare ffi::TypeInfo as a struct static member Eugene Zhulenev 2025-10-31 19:51:56 -0700
  • 8134117476 PR #32836: [GPU] Dispatch S-curve model to single-partition multi-host topology Terry Sun 2025-10-31 18:19:29 -0700
  • dad4fb74cd [xla:ffi] Remove deprecated TypeInfo constructor and replace it with XLA_FFI_TypeInfo alias Eugene Zhulenev 2025-10-31 17:58:56 -0700
  • fbd032df67 [xla:cpu] Pass HloModule pointer to Thunk SerDes Eugene Zhulenev 2025-10-31 16:51:10 -0700
  • 00be2bc09e [xla:cpu:onednn] Skip failing tests on Aarch64 CPUs. Penporn Koanantakool 2025-10-31 16:50:54 -0700
  • 0b5bc94a83 [xla:ffi] Migrate to xla::ffi::MakeTypeInfo() API Eugene Zhulenev 2025-10-31 16:24:48 -0700
  • 23f7b26bc5 Remove deprecated float_format/double_format in python proto text_format. Jie Luo 2025-10-31 16:21:11 -0700
  • 15e235f79b Allow IFRT-proxy to expand error-status payloads that are specific to Pathways. A. Unique TensorFlower 2025-10-31 15:47:38 -0700
  • c655468288 PR #31375: [XLA:GPU] Add NVLink domain check to CollectiveBackendAssigner Sevin Fide Varoglu 2025-10-31 15:24:38 -0700
  • bf84442f21 Refactor mesh and axis representation. Zixuan Jiang 2025-10-31 15:18:53 -0700
  • 9c620f90b8 [XLA][Numerics][HLO Original Value] Support original values for more cases in while loop simplifier pass Jian Cai 2025-10-31 15:10:43 -0700
  • a3f8740bc7 Update tflite schema to allow external buffer A. Unique TensorFlower 2025-10-31 14:58:17 -0700
  • 80048022c7 Update XNNPACK in XLA A. Unique TensorFlower 2025-10-31 14:18:53 -0700
  • 261e077984 [ReplicaGroupV3][MeshAxesReplicaGroupList][2/2] Add flattened_replica_groups function for MeshAxesReplicaGroupList. Bill Varcho 2025-10-31 14:00:07 -0700
  • a6e123761d [XLA][Numerics][HLO Original Values] Handles original values of while loops in TPU reduce code motion pass Jian Cai 2025-10-31 13:58:55 -0700
  • eef0661fc5 Rollforward with fixes of "Change RawSEDeviceMemory to be AsyncValueRef". Parker Schuh 2025-10-31 13:30:51 -0700
  • d008dc3999 Reverts d25ccb438d Bill Varcho 2025-10-31 12:19:30 -0700
  • 8572aaa4e9 Unify topology in PjRtTopologyDescription Haibo Huang 2025-10-31 12:13:20 -0700
  • e0f6a6c7f3 Integrate LLVM at llvm/llvm-project@42a8ff877d A. Unique TensorFlower 2025-10-31 11:53:16 -0700
  • 6ff7f9c87f Add de/serializaton of fake_allocations in DynamicSliceThunk. Aliia Khasanova 2025-10-31 10:30:18 -0700
  • ecc2510eb0 Use Deserializer lambda for embedded thunks in DynamicSliceThunk Eusebio Durán Montaña 2025-10-31 07:04:55 -0700
  • 26d0882419 Add proto serialization for GpuExecutable Henning Becker 2025-10-31 06:51:54 -0700
  • f73a954906 Add SymbolicExpr::IsBinaryOp() method A. Unique TensorFlower 2025-10-31 06:44:31 -0700
  • 718fe5695e [XLA:GPU] Add flags for filtering debugged thunks Marcin Radomski 2025-10-31 06:03:52 -0700
  • 4d78e8088a Automated Code Change A. Unique TensorFlower 2025-10-31 05:44:56 -0700
  • fe344908fa Automated Code Change A. Unique TensorFlower 2025-10-31 05:40:11 -0700
  • e7dcad735e Add equality operator for NamedSharding Kanish Anand 2025-10-31 05:07:03 -0700
  • add489fd8d Use std::vector<BufferAllocation> instead of std::vector<std::unique_ptr<BufferAllocation>> in DynamicSliceThunk. Aliia Khasanova 2025-10-31 04:58:35 -0700
  • 3326b0221f Automated Code Change A. Unique TensorFlower 2025-10-31 04:34:13 -0700
  • e32304ddc5 [Autotuner]Add support for sharded autotuning in the pass. A. Unique TensorFlower 2025-10-31 03:42:35 -0700
  • e32f20dd91 Use factory function to create CubSortThunk Eusebio Durán Montaña 2025-10-31 03:17:23 -0700
  • adfd891fde Refactor Mesh ctor's Kanish Anand 2025-10-31 03:15:28 -0700
  • 8734ec41d5 Disable capturing of dot RHS operands A. Unique TensorFlower 2025-10-31 02:22:48 -0700
  • d6d4e02248 [XLA:GPU] Add multimem setup. A. Unique TensorFlower 2025-10-31 02:10:52 -0700
  • a6e96588e3 compat: Update forward compatibility horizon to 2025-10-31 A. Unique TensorFlower 2025-10-31 02:04:05 -0700
  • f572aeee90 Update GraphDef version to 2397. A. Unique TensorFlower 2025-10-31 02:03:40 -0700
  • 993369077a Reverts bf23bf1b32 A. Unique TensorFlower 2025-10-31 01:39:41 -0700
  • d25ccb438d Reverts cef240807a A. Unique TensorFlower 2025-10-31 01:18:40 -0700
  • 4cfaa7e25c Automated Code Change A. Unique TensorFlower 2025-10-31 01:09:51 -0700
  • 5133f83425 Automated Code Change A. Unique TensorFlower 2025-10-31 00:40:43 -0700
  • ebacf2a211 Automated Code Change A. Unique TensorFlower 2025-10-30 23:30:33 -0700
  • cef240807a [ReplicaGroupV3][MeshAxesReplicaGroupList][1/2] Add initial class definition for V3 replica group. Bill Varcho 2025-10-30 22:59:12 -0700
  • d9c76aafeb Adjust the collective-permute cross host type to MULTI_HOST_NON_WORLD_LEVEL only. Felix Wang 2025-10-30 22:33:10 -0700
  • d90723f48e [xla:pjrt:cpu] Add e2e test for YnnFusion + PJRT client Eugene Zhulenev 2025-10-30 22:20:16 -0700
  • 7ad55e8818 [xla:cpu] Add an end-to-end test for ynn fusions Eugene Zhulenev 2025-10-30 22:01:05 -0700
  • bf23bf1b32 [xla:cpu] Pass HloModule pointer to Thunk SerDes Eugene Zhulenev 2025-10-30 21:37:36 -0700
  • 56d3b19280 [xla:cpu] NFC: Rename protos for Xnn/Ynn fusion options Eugene Zhulenev 2025-10-30 21:07:56 -0700
  • a95c558dc4 Save compile options with the compiled IFRT IR program to be used later for serialization A. Unique TensorFlower 2025-10-30 20:53:48 -0700
  • e61bac51b1 Automated Code Change A. Unique TensorFlower 2025-10-30 20:43:03 -0700
  • b2334ac330 Integrate LLVM at llvm/llvm-project@22079e3f36 A. Unique TensorFlower 2025-10-30 20:26:34 -0700
  • 6d86cff5f3 Automated Code Change A. Unique TensorFlower 2025-10-30 20:01:56 -0700
  • db273660ba [xla:pjrt] Remove PjRtFuture type alias Eugene Zhulenev 2025-10-30 19:34:12 -0700
  • 429a0cf1c7 [xla:cpu] Add target machine features to the error message Eugene Zhulenev 2025-10-30 17:42:32 -0700
  • d9024af6d4 [xla:cpu] Do not register legacy runtime symbols with XLA:CPU custom calls Eugene Zhulenev 2025-10-30 15:24:26 -0700
  • 31bb7c01ff Migrate multioutput_fusion_test to use PjRt. Niklas Vangerow 2025-10-30 15:09:38 -0700
  • c3d0bf7023 Add additional way to poision a connection (to allow testing different poisoning strategies). Parker Schuh 2025-10-30 14:41:12 -0700
  • c40bb10b96 Add the option to dump before/after autotuned instructions in AutotunerConfig. A. Unique TensorFlower 2025-10-30 13:24:39 -0700
  • 8f60516a86 Refactor: Move common SymbolicMapTest setup to the fixture. A. Unique TensorFlower 2025-10-30 13:23:41 -0700
  • 7736af79a6 Only enable YNNPACK for bf16 and int8 for now. A. Unique TensorFlower 2025-10-30 13:21:59 -0700
  • f4ebf9d47d [XLA][codegen] Migrate triton operations that have shared dialect lowerings are implemented for. Karlo Basioli 2025-10-30 13:17:07 -0700
  • 1424c4f739 Migrate slice_test to use PjRt. Niklas Vangerow 2025-10-30 13:15:41 -0700
  • 5973848600 [XLA][codegen] Emit shlo reshape from the fusion emitter and lower it to triton for the triton backend. Karlo Basioli 2025-10-30 12:48:50 -0700
  • f01a7fea8c Update ML Build Docker container to use hermetic C++ Quoc Truong 2025-10-30 12:48:50 -0700
  • 7e7b1a3015 Allow empty dimension list in SymbolicMap::ReplaceDimsAndSymbols A. Unique TensorFlower 2025-10-30 12:24:13 -0700
  • 175774337e Migrate params_test to use PjRt. Niklas Vangerow 2025-10-30 12:20:50 -0700
  • 5dcb571931 Remove unused code from convert.py Daniel Sosa 2025-10-30 12:20:27 -0700
  • 146c4f56b7 Clear frontend attributes for get-tuple-elements of GlobalToLocal and LocalToGlobal custom-calls. Zixuan Jiang 2025-10-30 12:12:21 -0700
  • a94890b1f9 Improve CUDNN error messages William S. Moses 2025-10-30 11:46:23 -0700
  • 9eeebc9be5 [XLA:GPU] Use a single intra-host ragged-all-to-all in the decomposition. Oleg Shyshkov 2025-10-30 11:43:00 -0700
  • 9b51864c7b [xla:ffi] Add example of async custom call in XLA:GPU Eugene Zhulenev 2025-10-30 11:39:42 -0700
  • 061041963e Migrate map_test to use PjRt. Niklas Vangerow 2025-10-30 11:06:56 -0700
  • dd3a14ace4 [Autotuner] Add sharding support using KeyValueStore Interface. A. Unique TensorFlower 2025-10-30 10:57:45 -0700
  • 4ffcba9004 [XLA][codegen] Emit stablehlo reduce op from the fusion emitter and lower it to triton for the triton backend. Karlo Basioli 2025-10-30 10:54:49 -0700
  • 0c87bef802 Migrate reshape_test to use PjRt. Niklas Vangerow 2025-10-30 10:19:16 -0700
  • 6dd75c4e8b [XTile] Modify Stable HLO check on iota to restrict it to the 1D case. Will Froom 2025-10-30 10:15:25 -0700
  • f2b36d1780 Integrate LLVM at llvm/llvm-project@4c46ae3948 A. Unique TensorFlower 2025-10-30 10:09:43 -0700
  • 3943b53326 Increase the maximum HLO op chain length for profiling from 8192 to 16384. Christian Sigg 2025-10-30 10:01:34 -0700
  • 71e640f242 Update Bazel version to 7.7.0. Yun Peng 2025-10-30 09:53:03 -0700
  • 1bef3e80b5 Reuse tuple elements field from existing HloSharding Kanish Anand 2025-10-30 09:09:15 -0700
  • d638f84b90 PR #33278: Bump keras from 3.11.3 to 3.12.0 in /xla/backends/cpu/benchmarks/e2e/gemma2/keras dependabot[bot] 2025-10-30 08:58:41 -0700
  • a28d4bf9f8 Add helper functions for creating and inspecting symbolic dimensions and symbols A. Unique TensorFlower 2025-10-30 08:57:41 -0700
  • 8d5e1015aa PR #4272: Qualcomm AI Engine Direct - Fix android-arm64 CMake build bug Graham 2025-10-30 08:34:56 -0700
  • a0921d9997 [XLA:GPU] CustomCallThunk: enable use of lambdas with captures Marcin Radomski 2025-10-30 08:30:30 -0700
  • 4461afa7ef [XLA:CPU] Compare host cpu features when loading AOT result to the compilation machine features Karlo Basioli 2025-10-30 08:28:39 -0700
  • fd85062199 Add hashing support for SymbolicMap A. Unique TensorFlower 2025-10-30 08:14:08 -0700
  • bec8916f32 [XLA:Collective] Remove unnecessary const for a function argument A. Unique TensorFlower 2025-10-30 08:09:28 -0700
  • 772ed8bbc7 Add serialization for ffi::Attribute Henning Becker 2025-10-30 08:07:43 -0700
  • 04fb26f2d1 [XLA:GPU] Add multi-host ragged-all-to-all decomposer pattern for combine ra2a. Oleg Shyshkov 2025-10-30 08:06:13 -0700
  • d69bbe355e Automated Code Change A. Unique TensorFlower 2025-10-30 07:46:47 -0700
  • 4df0c4afcd [XLA][codegen] Emit ReshapeToScalar as a tensor.extract op, and lower to triton specific impl. Karlo Basioli 2025-10-30 07:42:26 -0700
  • 345a251037 Add helper functions for creating constant SymbolicExpr A. Unique TensorFlower 2025-10-30 07:21:11 -0700
  • fd71e8be05 [XLA:GPU] Introduce CollectiveOpsE2ETestBase for common functionality. Oleg Shyshkov 2025-10-30 07:09:28 -0700