| PR |
0.60 |
Refactor Windows 2022 Dockerfile for better caching and secu |
|
Uses AI-style summary intro and bullet p |
2026-02-26 |
| COMMIT |
0.20 |
Automated Code Change |
|
Generic phrase 'Automated Code Change', |
2026-03-20 |
| COMMIT |
0.20 |
Automated Code Change |
|
Generic phrase 'Automated Code Change', |
2026-03-20 |
| COMMIT |
0.20 |
Automated Code Change |
|
Generic phrase 'Automated Code Change', |
2026-03-20 |
| PR |
0.20 |
Enforce strict bool type-checking for keepdims in reduce_* o |
|
Technical explanation and structured cha |
2026-03-22 |
| PR |
0.20 |
Fix idct() documentation: n parameter is already supported |
|
Technical doc update with regular engine |
2026-03-22 |
| PR |
0.15 |
Ensure quantized_dimension is zero in PrepareImpl for per-ch |
|
Slightly awkward language but technical |
2026-03-20 |
| PR |
0.10 |
Fix GIF decoder int32 overflow in frame offset pointer arith |
|
Technical and detailed, no AI-style phra |
2026-03-21 |
| PR |
0.10 |
Fix TFLite large_custom_options uint64 overflow bypassing bo |
|
Technical, domain-specific, no AI-style |
2026-03-21 |
| PR |
0.10 |
Fix BMP decoder integer overflow in pixel offset arithmetic |
|
Technical, domain-specific; no AI-style |
2026-03-21 |
| COMMIT |
0.05 |
PR #39500: [ROCm] Add rocm_rbe_dynamic config |
|
Some AI indicators (emojis, rigid sectio |
2026-03-20 |
| PR |
0.05 |
Fix int64 to int32 overflow in add_n scratch buffer allocati |
|
Technical, concise, domain-specific; no |
2026-03-22 |
| PR |
0.05 |
Fix oneDNN depthwise conv rank validation |
|
Terse, technical, includes changelog; no |
2026-03-22 |
| PR |
0.05 |
Fix CMake build issues for Android NDK r27d and Protobuf_PRO |
|
Brief, technical, specific build info; n |
2026-03-18 |
| PR |
0.05 |
docs: Fix FFT raw ops padding documentation |
|
Concise, technical fix summary; no AI si |
2026-03-20 |
| PR |
0.05 |
docs: Fix typo 'occured' to 'occurred' in metrics |
|
Template sections, typo fix, human tone; |
2026-03-21 |
| PR |
0.05 |
Add missing `&` in `reinterpret_cast` in `KeyedDenseTensorCo |
|
Brief, technical fix, specific reference |
2026-03-21 |
| PR |
0.01 |
Validate num_reduction_axes against XNN_MAX_TENSOR_DIMS in V |
|
Technical title and description; domain |
2026-03-23 |
| COMMIT |
0.00 |
Add SymbolicMap::GetSliceMap. |
|
Concise, technical; no AI phrasing detec |
2026-03-23 |
| COMMIT |
0.00 |
Expose cpu::TargetMachineOptions to XLA FFI. |
|
Direct technical explanation, informal t |
2026-03-23 |
| COMMIT |
0.00 |
Integrate LLVM at llvm/llvm-project@e1286d963eeb |
|
Terse LLVM update, domain specific wordi |
2026-03-23 |
| COMMIT |
0.00 |
Remove test timeout long from tests that don't need it. |
|
Brief justification, informal, domain-sp |
2026-03-23 |
| COMMIT |
0.00 |
[XLA:GPU] CudaBlasLtTest: fully initialize MemoryArgs |
|
Short, technical, uses domain abbreviati |
2026-03-23 |
| COMMIT |
0.00 |
Refactor FFI scratch size estimation and update CubScan hand |
|
Technical explanation, slight formal phr |
2026-03-23 |
| COMMIT |
0.00 |
[XLA:GPU] Use collective allocator for ragged-all-to-all bar |
|
Technical and concise, jargon present. |
2026-03-23 |
| COMMIT |
0.00 |
Remove test timeout long from iota_test. |
|
Slightly more verbose; still technical a |
2026-03-23 |
| COMMIT |
0.00 |
Remove unnecessary parentheses from SymbolicExpr and Symboli |
|
Casual justification, technical referenc |
2026-03-23 |
| COMMIT |
0.00 |
[XLA:GPU] Remove unused header. |
|
Very brief, direct, technical. |
2026-03-23 |
| COMMIT |
0.00 |
Factor out the conflict resolution logic for call argument a |
|
Technical refactoring phrasing; no AI si |
2026-03-23 |
| COMMIT |
0.00 |
Rollback due to presubmit test breakage. |
|
Terse rollback description; domain-speci |
2026-03-23 |
| COMMIT |
0.00 |
PR #38613: [ROCM] amdgpu_backend refactoring and new HSACO c |
|
Technical summary; domain-specific abbre |
2026-03-23 |
| COMMIT |
0.00 |
[XLA:GPU] Use adjacency lists in PrimitiveUtilTest.CastPrese |
|
Brief technical commit; uses domain jarg |
2026-03-23 |
| COMMIT |
0.00 |
Delete now unused and duplicate unit test on shardy outliner |
|
Direct statement for test deletion; conc |
2026-03-23 |
| COMMIT |
0.00 |
Reverts 8a7b027af15a56ac3ebe69cb57dff9157c7264c2 |
|
Standard commit style for revert; domain |
2026-03-23 |
| COMMIT |
0.00 |
[XLA:CPU]: Make OneAPI builds non blocking |
|
Jargon and concise; typical technical co |
2026-03-23 |
| COMMIT |
0.00 |
Fix uninitialized bool use_async_dispatch in GetStreamExecut |
|
Detailed technical explanation; no AI ph |
2026-03-23 |
| COMMIT |
0.00 |
[XLA] Disalbe LICM on scalar and support Range calculation f |
|
Typo present; terse and informal, human |
2026-03-22 |
| COMMIT |
0.00 |
Get stablehlo import as a wrapper over sdy import to test pr |
|
Informal, terse, and uses domain-specifi |
2026-03-21 |
| COMMIT |
0.00 |
Unsupport dedup-functions-fully on unflattenner properly. |
|
Brief, technical, with domain terms and |
2026-03-21 |
| COMMIT |
0.00 |
Reverts 05e79d3120a0413dd6ecf9463e40b1e7227f52cb |
|
Standard revert message; no AI hallmarks |
2026-03-21 |
| COMMIT |
0.00 |
Rollback due to internal test breakage |
|
Terse, domain-specific, typical commit m |
2026-03-21 |
| COMMIT |
0.00 |
Do not ignore memory space validation in strict shape checki |
|
Technical, concise, no AI signal. |
2026-03-21 |
| COMMIT |
0.00 |
Integrate LLVM at llvm/llvm-project@a70419505471 |
|
Concise, repo-maintenance language, huma |
2026-03-20 |
| COMMIT |
0.00 |
Create absl::string_view versions of some TSL functions. |
|
Contains rationale and context, human to |
2026-03-20 |
| COMMIT |
0.00 |
Refactor: Use range-based for loops in SPMD handlers. |
|
Direct, domain jargon, standard engineer |
2026-03-20 |
| COMMIT |
0.00 |
Update NCCL version to 2.28.9 in Tensorflow. |
|
Succinct and typical for such an update. |
2026-03-20 |
| COMMIT |
0.00 |
[XLA:GPU] Add MultiGpuBarrierWithNcclKernel. |
|
Technical, slightly imperfect, human-wri |
2026-03-20 |
| COMMIT |
0.00 |
PR #39562: [ROCm] Set jobs timeout, preventing blocking node |
|
Template elements, but minor free-text a |
2026-03-20 |
| COMMIT |
0.00 |
PR #39453: [xla:gpu] Use default compute stream for compute |
|
Domain-specific, concise, natural phrasi |
2026-03-20 |
| COMMIT |
0.00 |
Factor out the conflict resolution logic for call argument a |
|
Technical explanation, minor imperfectio |
2026-03-20 |
| COMMIT |
0.00 |
PR #39287: [ROCm] Add CUB support for scan (prefix sum) op |
|
Technical and detailed, but domain-speci |
2026-03-20 |
| COMMIT |
0.00 |
[IFRT IR] Modify merge reshards pass to handle both copies a |
|
Terse commit style, no signals of AI phr |
2026-03-20 |
| COMMIT |
0.00 |
Add function for determining compatibility of two GpuTopolog |
|
Direct domain language, typos show human |
2026-03-20 |
| COMMIT |
0.00 |
[Convolutions] Unify ConvKind enum with ConvolutionKind from |
|
Terse, technical description, not AI-lik |
2026-03-20 |
| COMMIT |
0.00 |
PR #39466: [XLA:GPU][oneAPI] Add oneAPI GPU runner property |
|
Domain-specific, copier boilerplate is e |
2026-03-20 |
| COMMIT |
0.00 |
PR #39526: [XLA:GPU][oneAPI] Disable int4 related tests on I |
|
Brief, technical, no AI stylistic hallma |
2026-03-20 |
| COMMIT |
0.00 |
Delete unused MLIR pybindings/model utils setup in TF repo |
|
Very terse with no free-text, certainly |
2026-03-20 |
| COMMIT |
0.00 |
PR #39405: Bump numpy from 2.4.2 to 2.4.3 |
|
— |
2026-03-20 |
| COMMIT |
0.00 |
Handle parameter sharding conflict resolution in shardy. |
|
Very terse, technical and domain-specifi |
2026-03-20 |
| COMMIT |
0.00 |
In MultiDimBroadcast tests, create the model once and resize |
|
Casual tone, explanatory but no AI indic |
2026-03-20 |
| COMMIT |
0.00 |
Fix crash when decoding "upside down" .bmp files |
|
Concise commit message with technical ja |
2026-03-20 |
| COMMIT |
0.00 |
[XLA] TupleSimplifier - only update schedule if changes were |
|
Technical and terse message, typical hum |
2026-03-20 |
| COMMIT |
0.00 |
Build: Bump ubuntu 22.04 base image |
|
Straightforward commit message; slight f |
2026-03-20 |
| COMMIT |
0.00 |
[VHLO-PJRT] Only allow certain known-stable dialects in mixe |
|
Terse, domain-specific, standard commit |
2026-03-20 |
| COMMIT |
0.00 |
Make IfrtBackendCompiler own a tpu compiler. |
|
Technical, concise, clear commit message |
2026-03-20 |
| COMMIT |
0.00 |
Remove the local clamp in `HandleGatherWithoutConflicts` and |
|
Detailed, domain-specific reasoning, hum |
2026-03-20 |
| COMMIT |
0.00 |
Support `alias_data` in the TFLite XNNPack delegate weight c |
|
Technical jargon, succinct phrasing, hum |
2026-03-20 |
| COMMIT |
0.00 |
[XLA:GPU] FusionPipeline: remove HloPassFix<> |
|
Informal, domain-specific explanation, h |
2026-03-20 |
| COMMIT |
0.00 |
Automated Code Change |
|
Standard human template, no AI hallmarks |
2026-03-20 |
| COMMIT |
0.00 |
[CublasLt] Allow S8xS8ToS32 cuBLASLt gemms in XLA GPU. |
|
Technical abbreviation, terse style, hum |
2026-03-20 |
| COMMIT |
0.00 |
[XLA:GPU] Move SplitK pass (disabled by default) to right be |
|
Domain-specific explanation, normal comm |
2026-03-20 |
| COMMIT |
0.00 |
Integrate Triton up to [443f4fe96e](https://github.com/trito |
|
Structured integration log, no AI signal |
2026-03-20 |
| COMMIT |
0.00 |
Automated Code Change |
|
Standard human template, no AI hallmarks |
2026-03-20 |
| COMMIT |
0.00 |
tsl: use Absl synchronization primitives |
|
Human explanation, domain-specific terms |
2026-03-20 |
| COMMIT |
0.00 |
[XLA] Backend config on disk inlining: schema & read support |
|
Technical language and concise explanati |
2026-03-20 |
| COMMIT |
0.00 |
[XLA:GPU] Emit cwise ops with the new tiling. |
|
Brief, domain-specific, minimal prose; h |
2026-03-20 |
| COMMIT |
0.00 |
Add several regression tests for LocalRendezvous |
|
Casual, technical tone and domain detail |
2026-03-20 |
| COMMIT |
0.00 |
[XLA:Runtime]: Remove unused method from buffer_sequencing_e |
|
Technical reasoning, informal phrase; na |
2026-03-20 |
| COMMIT |
0.00 |
Avoid logging that a test is being skipped repeatedly in eve |
|
Direct, minimal, domain-specific context |
2026-03-20 |
| COMMIT |
0.00 |
Add support for GB300, B300 and fix bug for GB200. |
|
Abbreviations, technical tone, no AI sig |
2026-03-20 |
| COMMIT |
0.00 |
Add more compilation logs for ReduceScatter decisions |
|
Technical, informal, minimal; clearly hu |
2026-03-20 |
| COMMIT |
0.00 |
Remove `TileAssignment::UsesDevice`. |
|
Technical language, terse, no AI signs. |
2026-03-20 |
| COMMIT |
0.00 |
Reverts 8ea661247f17f05467efd77110a63206d725df8a |
|
Short revert message, standard for commi |
2026-03-20 |
| COMMIT |
0.00 |
Set default xla_gpu_experimental_ragged_all_to_all_use_barri |
|
Casual phrasing, domain-specific, human |
2026-03-19 |
| COMMIT |
0.00 |
Explain FunctionDef parsing failures in MLIR bridge pass. |
|
Technical explanation with domain contex |
2026-03-19 |
| COMMIT |
0.00 |
Add documentation for the `isolate_hlo` tool |
|
Brief, no formal phrasing, human. |
2026-03-19 |
| COMMIT |
0.00 |
Integrate StableHLO at openxla/stablehlo@1373ee06 |
|
Integration message, no AI signal, human |
2026-03-19 |
| COMMIT |
0.00 |
Update copyright year in isolate_hlo_test.sh. |
|
Simple update, terse, human-written. |
2026-03-19 |
| COMMIT |
0.00 |
Add TF_CALL_float8_e4m3fn(CASE) to batch related ops |
|
Technical macro addition, human text sty |
2026-03-19 |
| COMMIT |
0.00 |
[IFRT IR] Rename ir_py serde methods to better capture that |
|
Domain abbreviations, terse change expla |
2026-03-19 |
| COMMIT |
0.00 |
[IFRT IR] Separate AddControlDependencies pass to avoid runn |
|
Clear technical change, no formal AI sty |
2026-03-19 |
| COMMIT |
0.00 |
Prioritize meshes with more **non-size-one axes** during ded |
|
Technical language, domain-specific, cas |
2026-03-19 |
| COMMIT |
0.00 |
Integrate LLVM at llvm/llvm-project@b03b58be38f9 |
|
Plain, minimal update; human style, no A |
2026-03-19 |
| COMMIT |
0.00 |
Add overwrite option for TSL file rename. |
|
Direct technical explanation, terse and |
2026-03-19 |
| COMMIT |
0.00 |
[xla:cpu:ynn] Make LibraryRewriter support `LIBRARY_FUSION_T |
|
Informal, domain-heavy, incomplete sente |
2026-03-19 |
| COMMIT |
0.00 |
Collective combiner: skip reachable instructions instead of |
|
Brief commit message, human style, no AI |
2026-03-19 |
| COMMIT |
0.00 |
Remove dead code. tiling_dim_factors is never used in Reshar |
|
Very terse, domain-specific; typical hum |
2026-03-19 |
| COMMIT |
0.00 |
PR #39501: [tsl] Add tsl::TiedAny for tying lifetime of any |
|
Technical jargon, changelog, no AI phras |
2026-03-19 |
| COMMIT |
0.00 |
Rollback of PR #39386 |
|
Rollback message, standard human commit |
2026-03-19 |
| COMMIT |
0.00 |
[PjRt-IFRT] Add a parameter memory kind field in SerializedX |
|
Technical update, domain language, no AI |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Add support for packing RegisteredBufferHandle as |
|
Concise, domain-specific, no AI traits p |
2026-03-19 |
| COMMIT |
0.00 |
[tsl] Make base64.h use absl::Status and absl::string_view |
|
Concise, terse commit; domain-specific j |
2026-03-19 |
| COMMIT |
0.00 |
Migrate reduce_window_test_library and tests to PjRt runner. |
|
Short, direct, human-written engineering |
2026-03-19 |
| COMMIT |
0.00 |
Enable experimental Triton scaled dot product by default. |
|
Direct technical summary; normal human s |
2026-03-19 |
| COMMIT |
0.00 |
PR #39105: [GPU] Fix autotuning of GEMM fusions with cuDNN b |
|
Template structure; summary/justificatio |
2026-03-19 |
| COMMIT |
0.00 |
Unflatten the call graph after shardy outlining. |
|
Casual explanation with domain-specific |
2026-03-19 |
| COMMIT |
0.00 |
PR #39386: Fix use-after-free of temporary Literal in Linear |
|
Technical bug explanation; details sugge |
2026-03-19 |
| COMMIT |
0.00 |
PR #39413: Add fusion debug logging |
|
Debug logging details; summary is mildly |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU]: Add a host callback registry to monitor CUDA stre |
|
Terse, technical change description, hum |
2026-03-19 |
| COMMIT |
0.00 |
Add SymbolicMap::Evaluate method. |
|
Brief, domain-specific commit message. |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Skip non-fusible computations in GemmFusion and Ge |
|
Very terse, template-style commit messag |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Remove a dead function from target_util |
|
Concise technical phrasing, domain-speci |
2026-03-19 |
| COMMIT |
0.00 |
Automated Code Change |
|
Generic wording but normal for automated |
2026-03-19 |
| COMMIT |
0.00 |
PR #39469: [tsl] Add `TiedRef` and `Tied` lifetime-tyi |
|
Technical jargon, natural changelog, no |
2026-03-19 |
| PR |
0.00 |
[XLA:GPU] Add Tie method to GpuClique and CollectiveCliques. |
|
— |
2026-03-23 |
| PR |
0.00 |
Add SymbolicMap::GetSliceMap. |
|
— |
2026-03-19 |
| PR |
0.00 |
Refactor to first pick functions and then apply on calls. |
|
— |
2026-03-20 |
| PR |
0.00 |
Build: Upgrade base image packages |
|
— |
2026-03-23 |
| PR |
0.00 |
Simplify reduce_window_test setup. |
|
— |
2026-03-23 |
| PR |
0.00 |
Expose cpu::TargetMachineOptions to XLA FFI. |
|
— |
2026-03-19 |
| PR |
0.00 |
Store xla::cpu::TargetMachineOptions in GpuExecutable and pa |
|
— |
2026-03-23 |
| PR |
0.00 |
WIP: ML-Actions cont'd |
|
— |
2026-03-21 |
| PR |
0.00 |
Integrate LLVM at llvm/llvm-project@e1286d963eeb |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU] Implement round-nearest-even support and enable BF |
|
— |
2026-03-17 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Remove test timeout long from tests that don't need it. |
|
— |
2026-03-23 |
| PR |
0.00 |
PR #38909: [ROCm] Support hipblaslt group-gemm 3/5 |
|
— |
2026-03-17 |
| PR |
0.00 |
PR #39530: [XLA:GPU][oneAPI][Bug fix] Add autotune cache key |
|
— |
2026-03-23 |
| PR |
0.00 |
PR #39607: [ROCm] Handle cancellation in multi_gpu step of r |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU] lit_device_test build rule |
|
— |
2026-03-19 |
| PR |
0.00 |
PR #39021: [se:cuda] Add MemoryAllocators for all memory kin |
|
— |
2026-03-17 |
| PR |
0.00 |
[XLA:GPU] CudaBlasLtTest: fully initialize MemoryArgs |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Refactor FFI scratch size estimation and update CubScan hand |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU]: Unify the host callback registry across all strea |
|
— |
2026-03-23 |
| PR |
0.00 |
Support single replica - multiple infeed in HLORunner |
|
— |
2025-09-30 |
| PR |
0.00 |
[XLA:GPU] Make GpuPerformanceModelCache thread-safe |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU] Use collective allocator for ragged-all-to-all bar |
|
— |
2026-03-20 |
| PR |
0.00 |
Unflatten call graph by deduping functions fully and picking |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Remove test timeout long from iota_test. |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Remove unnecessary parentheses from SymbolicExpr and Symboli |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
mu upstream binding-less core api |
|
— |
2025-05-24 |
| PR |
0.00 |
[XLA:GPU] Remove unused header. |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Factor out the conflict resolution logic for call argument a |
|
— |
2026-03-21 |
| PR |
0.00 |
[CublasLt] Enable CublasLt on default. |
|
— |
2026-03-23 |
| PR |
0.00 |
Use nanobind for tfl converter apis |
|
— |
2025-04-07 |
| PR |
0.00 |
Rollback due to presubmit test breakage. |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
PR #38613: [ROCM] amdgpu_backend refactoring and new HSACO c |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA:GPU] Use adjacency lists in PrimitiveUtilTest.CastPrese |
|
— |
2026-03-19 |
| PR |
0.00 |
Push shardy outliner up past StablehloRoundTripShardMapExpor |
|
— |
2026-03-23 |
| PR |
0.00 |
Delete now unused and duplicate unit test on shardy outliner |
|
— |
2026-03-21 |
| PR |
0.00 |
Split outliner and unflattener by completely delegating dedu |
|
— |
2026-03-21 |
| PR |
0.00 |
[XLA:GPU] Update caches in MultiOutputFusion pass. |
|
— |
2026-03-20 |
| PR |
0.00 |
Store xla::cpu::TargetMachineOptions in GpuExecutable and pa |
|
— |
2026-03-23 |
| PR |
0.00 |
Reverts 8a7b027af15a56ac3ebe69cb57dff9157c7264c2 |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Delegate deduplicating fully to unflattener and drop it from |
|
— |
2026-03-21 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
[XLA:CPU]: Make OneAPI builds non blocking |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA] Add utilities to read backend config (for 1P/3P users) |
|
— |
2026-03-23 |
| PR |
0.00 |
Disable tf-reduce-identity.mlir test for open source. |
|
— |
2026-03-23 |
| PR |
0.00 |
Add aggregation mode to cupti collector. |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-18 |
| PR |
0.00 |
Add optional warmup with profiler to MultihostHloRunner. |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Internal only: experimental |
|
— |
2025-04-03 |
| PR |
0.00 |
Rewrite HloScanInstruction to tree reduction in ReduceWindow |
|
— |
2026-03-16 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU]: Use the host callback registry for cuda stream. |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
[XLA:GPU] Restore cusolver rpath in all_runtime |
|
Contains domain-specific context and inf |
2026-03-22 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Do not call `HostShapeToDeviceShape` on device shapes |
|
— |
2026-03-21 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-23 |
| PR |
0.00 |
Fix uninitialized bool use_async_dispatch in GetStreamExecut |
|
— |
2026-03-22 |
| PR |
0.00 |
Add utility for constructing a basic CommonPjRtLoadedExecuta |
|
— |
2026-03-19 |
| PR |
0.00 |
Import samsung option changes |
|
— |
2026-03-10 |
| PR |
0.00 |
Support the XLA GPU compilation flags in Orbax |
|
— |
2026-03-23 |
