| PR |
1.00 |
Fix kaiser_window for negative beta values |
|
Signature states 'Generated with Claude |
2026-03-18 |
| PR |
1.00 |
Fix tf.unravel_index to reject negative indices |
|
Explicitly states 'Generated with Claude |
2026-03-18 |
| PR |
1.00 |
Fix crash on negative ConfigProto thread counts |
|
Explicit AI signature: 'Generated with C |
2026-03-18 |
| PR |
1.00 |
Fix tf.io.encode_png crash on zero-dimension tensors |
|
Explicit AI signature: 'Generated with C |
2026-03-18 |
| PR |
1.00 |
Validate grad tensor rank in AvgPool3DGrad shape function |
|
Explicit AI attribution: 'Generated with |
2026-03-18 |
| PR |
0.25 |
docs: Fix FFT raw ops padding documentation |
|
Verbose but technical; some mild over-ex |
2026-03-19 |
| 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 |
0.15 |
Fix IRFFT crash on single-element complex input |
|
Mostly technical with minor verbosity, l |
2026-03-18 |
| PR |
0.15 |
Fix SIGSEGV from integer overflow in TFLite tensor dimension |
|
Technical changelog, domain-specific, no |
2026-03-18 |
| PR |
0.12 |
Fix model.fit() crash with unknown TensorShape from tf.numpy |
|
Uses GitHub issue, domain jargon, concis |
2026-03-18 |
| PR |
0.10 |
Fix numerical instability in tf.norm and EuclideanNorm gradi |
|
Technical, detailed; domain-aligned styl |
2026-03-18 |
| PR |
0.10 |
Fix conv1d eager/graph mode inconsistency for zero-sized out |
|
Technical, practical, minor truncation s |
2026-03-18 |
| PR |
0.10 |
Fix GPU MatrixSolve singular matrix detection for all shapes |
|
Detailed technical explanation, human to |
2026-03-18 |
| PR |
0.08 |
Fix broken padding documentation links in nn_ops and nn_impl |
|
Doc maintenance, specific details, infor |
2026-03-18 |
| PR |
0.05 |
docs: Fix FFT raw ops padding documentation |
|
Concise, technical, natural; no AI hallm |
2026-03-20 |
| PR |
0.05 |
Ensure quantized_dimension is zero in PrepareImpl for per-ch |
|
Technical description with colloquial to |
2026-03-20 |
| PR |
0.05 |
Validate negative parallelism thread counts in Session |
|
Technical, concise, domain-specific; no |
2026-03-18 |
| PR |
0.05 |
Fix integer overflow in Tile kernel MultiplyShapeDims |
|
Technical explanation with security focu |
2026-03-18 |
| PR |
0.05 |
Fix MKL Conv3D/Conv2D crash on non-matching input rank |
|
Concise domain language; no AI hallmarks |
2026-03-18 |
| PR |
0.05 |
Fix cumulative_logsumexp NaN propagation |
|
Technical explanations, human tone, doma |
2026-03-18 |
| PR |
0.05 |
Fix MaxPool3D MKL/oneDNN CHECK failure on non-5D input |
|
Technical and concise, filled in templat |
2026-03-18 |
| PR |
0.05 |
Fix TensorArray.split silently accepting mismatched lengths |
|
Brief, technical, domain-driven; not AI- |
2026-03-18 |
| PR |
0.05 |
Fix integer overflow in Detection PostProcess (heap buffer o |
|
Security-focused, domain language, human |
2026-03-18 |
| PR |
0.05 |
Fix AvgPool3DGrad CHECK failure on invalid grad rank in MKL |
|
Detailed technical changelog, domain-spe |
2026-03-18 |
| PR |
0.05 |
Fix tf.linalg.logdet returning NaN for non-SPD matrices |
|
Domain-specific jargon, thorough technic |
2026-03-18 |
| PR |
0.05 |
Fix CMake build issues for Android NDK r27d and Protobuf_PRO |
|
Direct, technical, informal phrasing; sc |
2026-03-18 |
| 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 |
| COMMIT |
0.00 |
Update conditions in ```CanReshardWithCollectivePermute``` t |
|
Short, domain-specific update, human com |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:Collective] Support more cases in collective pipeliner. |
|
Technical, lists changes in terse style, |
2026-03-19 |
| COMMIT |
0.00 |
[XLA] Add MoveSingleElement function to the permutation util |
|
Brief, domain jargon, natural commit phr |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Refactor NcclCommunicator::RegisterBufferOnce to r |
|
Technical refactor, concise, no AI signa |
2026-03-19 |
| COMMIT |
0.00 |
Add registered codegen backends to KV store key, for multi-h |
|
Elaborate explanation but technical, lac |
2026-03-19 |
| COMMIT |
0.00 |
Automated Code Change |
|
Generic automated change, possible scrip |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Update SplitK pass heuristic values to the Februar |
|
Technical, concise, date and domain cont |
2026-03-19 |
| COMMIT |
0.00 |
Enable Modulo 1 simplification in SymbolicExpr. |
|
Concise technical explanation with infor |
2026-03-19 |
| COMMIT |
0.00 |
[xla:gpu] NFC: Clean up CUB sort and scan FFI custom call ta |
|
Terse changelog, domain-specific, likely |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU/CPU] Disable MLIR pass verifier in optimized builds |
|
Clear technical summary, no AI phrasing |
2026-03-19 |
| COMMIT |
0.00 |
[XLA:GPU] Rename functions of the DotOperandDims helper. |
|
Informal, human style ('Hopefully'), min |
2026-03-19 |
| COMMIT |
0.00 |
Add simplification for (x % a) % b when a % b == 0. |
|
Technical jargon, straightforward summar |
2026-03-19 |
| COMMIT |
0.00 |
[xla:gpu] Various fixes for lowering the scan op to a FFI cu |
|
Technical detail, minor cutoff at end, h |
2026-03-19 |
| COMMIT |
0.00 |
Add a fallback for `PjRtCApiTopologyDescription::Fingerprint |
|
Fallback changelog, no AI signals, very |
2026-03-19 |
| COMMIT |
0.00 |
Statically register XLA:GPU's PJRT client |
|
Technical registration summary, no AI tr |
2026-03-19 |
| COMMIT |
0.00 |
[Tiling] Update indexing performance model to support region |
|
Domain jargon, concise revert notice, hu |
2026-03-19 |
| COMMIT |
0.00 |
Update copyright year in isolate_hlo.cc. |
|
Short copyright update, minimal human ed |
2026-03-19 |
| COMMIT |
0.00 |
Automated Code Change |
|
Minimal phrasing, normal commit automati |
2026-03-19 |
| COMMIT |
0.00 |
Update TSL `FileExists` to use `absl::string_view`. |
|
Concise technical language, domain vocab |
2026-03-19 |
| COMMIT |
0.00 |
Support passing payload from absl::Status in CreateErrorBuff |
|
Short, terse, domain-specific, clear hum |
2026-03-19 |
| COMMIT |
0.00 |
[PjRt] Add PjRt(Loaded)Executable::GetParameterMemoryKinds() |
|
Structured, technical explanation, no AI |
2026-03-19 |
| COMMIT |
0.00 |
change the batch_size attribute to max_valency attribute |
|
Brief, technical terminology, human-like |
2026-03-19 |
| COMMIT |
0.00 |
Generalize InferDispatchInfo to handle the construction of |
|
Jargon and specific context, no AI-typic |
2026-03-19 |
| COMMIT |
0.00 |
Handle replicated sharding in `HloSharding::dimension`. |
|
Casual technical tone, domain-specific d |
2026-03-19 |
| COMMIT |
0.00 |
Add `HandleGatherWithoutConflicts`, which is a short cut to |
|
Domain detail, structured explanation, l |
2026-03-19 |
| COMMIT |
0.00 |
Remove deprecated `IsTileMaximal` and `IsTileMaximalLeaf`, w |
|
Direct release note style, technical ter |
2026-03-19 |
| COMMIT |
0.00 |
Expose GetInterfaceAddressesHelper via MegaScale PJRT extens |
|
Brief, technical, domain-specific, no AI |
2026-03-19 |
| COMMIT |
0.00 |
cc_embed_data: Properly strip directory prefixes. |
|
Terse domain-specific phrasing, no AI si |
2026-03-19 |
| COMMIT |
0.00 |
Rename `num_warmup_threads` to `num_warmup_batch_threads` in |
|
Direct, technical language, typical huma |
2026-03-18 |
| COMMIT |
0.00 |
Extract cublas(lt) along with get-tuple during autotuning. |
|
Contains jargon and casual tone, human-w |
2026-03-18 |
| COMMIT |
0.00 |
[XLA:CPU] Rewrite thunk_serders_dynamic_tests as one paramet |
|
Short and technical; no AI hallmarks. |
2026-03-18 |
| COMMIT |
0.00 |
Remove legacy serialization as 12w compat window has passed |
|
Uses domain references and concise phras |
2026-03-18 |
| COMMIT |
0.00 |
memfile_builtin: Disallow empty FileToc names. |
|
Brief, jargon-rich; lacks AI-style boile |
2026-03-18 |
| COMMIT |
0.00 |
Integrate StableHLO at openxla/stablehlo@78429ebf |
|
Standard update description; human techn |
2026-03-18 |
| COMMIT |
0.00 |
[ReplicaGroupV3] Add support for parsing MeshAxesReplicaGrou |
|
Uses technical terms, no signs of AI tex |
2026-03-18 |
| COMMIT |
0.00 |
When a weight is cached by the XNNPack delegate, mark origin |
|
Casual, domain-specific; typical human s |
2026-03-18 |
| COMMIT |
0.00 |
[XLA:CPU] Implement CustomCallThunk serdes. |
|
Technical, concise; lacks AI-generated s |
2026-03-18 |
| COMMIT |
0.00 |
Disable HLO Sharding V3 in V1/V2 SPMD partitioner tests. |
|
Concise technical description, human eng |
2026-03-18 |
| COMMIT |
0.00 |
Avoid casting dimensions to float in STRIDED_SLICE ResizeOut |
|
Uses technical jargon and domain-specifi |
2026-03-18 |
| COMMIT |
0.00 |
[IFRT IR] Do not enable HloShardingV3 at lowering time |
|
Very brief, domain-specific language. |
2026-03-18 |
| COMMIT |
0.00 |
Add ABSL_CONST_INIT to kCheckpointCallbackManagerResourceNam |
|
Single technical line, typical for human |
2026-03-18 |
| COMMIT |
0.00 |
PyClient::CompileAndLoad is already cloning the mlir modules |
|
Contains domain-specific references, ter |
2026-03-18 |
| COMMIT |
0.00 |
ifrt_proxy_contrib_pathways: Change function to accept `absl |
|
Technical change described succinctly, h |
2026-03-18 |
| COMMIT |
0.00 |
PR #39014: [xla:pjrt] Unify AsyncWorkRunner API with tsl::Ex |
|
Structured changelog, project-specific a |
2026-03-18 |
| COMMIT |
0.00 |
PR #38919: [ROCm] CI: Add GHA workflow for ROCm JAX unit tes |
|
Template sections, domain abbreviations, |
2026-03-18 |
| COMMIT |
0.00 |
[XLA:GPU] hlo_to_xtileir tool for lit |
|
Short, domain-specific term, not AI-like |
2026-03-18 |
| COMMIT |
0.00 |
PR #39279: [ROCm] CI: Fix containers settings |
|
Terse, technical summary with clear huma |
2026-03-18 |
| PR |
0.00 |
[XLA:GPU] Add MultiGpuBarrierWithNcclKernel. |
|
— |
2026-03-20 |
| PR |
0.00 |
WIP: ML-Actions GHA workflows |
|
— |
2026-03-19 |
| PR |
0.00 |
Handle parameter sharding conflict resolution in shardy. |
|
— |
2026-03-19 |
| PR |
0.00 |
In MultiDimBroadcast tests, create the model once and resize |
|
— |
2026-03-18 |
| PR |
0.00 |
[XLA] TupleSimplifier - only update schedule if changes were |
|
— |
2026-03-20 |
| PR |
0.00 |
[VHLO-PJRT] Only allow certain known-stable dialects in mixe |
|
— |
2026-03-17 |
| PR |
0.00 |
Factor out the conflict resolution logic for call argument a |
|
— |
2026-03-19 |
| PR |
0.00 |
Update resharding logic to use HasPartialReplication and do |
|
— |
2026-03-20 |
| PR |
0.00 |
Use debug strings rather than bytes for the un-matched XlaSh |
|
— |
2026-03-20 |
| PR |
0.00 |
Add method to get implicitly replicated axes in ``NamedShard |
|
— |
2026-03-20 |
| PR |
0.00 |
Fix crash when parsing "upside down" .bmp files |
|
— |
2026-03-20 |
| PR |
0.00 |
Build: Bump ubuntu 22.04 base image |
|
— |
2026-03-18 |
| PR |
0.00 |
Make IfrtBackendCompiler own a tpu compiler. |
|
— |
2026-03-19 |
| PR |
0.00 |
Add aggregation mode to cupti collector. |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-01-09 |
| PR |
0.00 |
Remove the local clamp in `HandleGatherWithoutConflicts` and |
|
— |
2026-03-19 |
| PR |
0.00 |
PR #39562: [ROCm] Set jobs timeout, preventing blocking node |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-15 |
| PR |
0.00 |
Support `alias_data` in the TFLite XNNPack delegate weight c |
|
— |
2026-03-13 |
| PR |
0.00 |
[XLA:GPU] Update caches in MultiOutputFusion pass. |
|
— |
2026-03-20 |
| PR |
0.00 |
Add function for determining compatibility of two GpuTopolog |
|
— |
2026-03-20 |
| PR |
0.00 |
Testing changes only. |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA:GPU] FusionPipeline: remove HloPassFix<> |
|
— |
2026-03-19 |
| PR |
0.00 |
Replicate the scalar scaling factor for 4-bit tensor-wise qu |
|
— |
2026-03-18 |
| PR |
0.00 |
[XLA:GPU] lit_device_test build rule |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA:GPU] Use adjacency lists in PrimitiveUtilTest.CastPrese |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-14 |
| PR |
0.00 |
PR #39287: [ROCm] Add CUB support for scan (prefix sum) op |
|
— |
2026-03-19 |
| PR |
0.00 |
[CublasLt] Allow S8xS8ToS32 cuBLASLt gemms in XLA GPU. |
|
— |
2026-03-17 |
| PR |
0.00 |
[XLA:GPU]: Use error callbacks in the runtime to notify of a |
|
— |
2026-03-19 |
| PR |
0.00 |
[XLA:GPU] Move SplitK pass (disabled by default) to right be |
|
— |
2026-03-19 |
| PR |
0.00 |
[XLA:GPU]: Use the host callback registry for cuda stream. |
|
— |
2026-03-19 |
| PR |
0.00 |
Integrate Triton up to [443f4fe96e](https://github.com/trito |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Plumb the `num_warmup_batch_threads` BatchFunction op attrib |
|
— |
2026-03-20 |
| PR |
0.00 |
Add num_warmup_batch_threads as a batching op attribute. |
|
— |
2026-03-20 |
| PR |
0.00 |
PR #39405: Bump numpy from 2.4.2 to 2.4.3 |
|
— |
2026-03-20 |
| PR |
0.00 |
tsl: use Absl synchronization primitives |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
PR #39508: [XLA:GPU][oneAPI] Remove hardcoded spirv-binary i |
|
— |
2026-03-20 |
| PR |
0.00 |
PR #39466: [XLA:GPU][oneAPI] Add oneAPI GPU runner property |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA] Backend config on disk inlining: schema & read support |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA:GPU] Emit cwise ops with the new tiling. |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Add several regression tests for LocalRendezvous |
|
— |
2026-03-18 |
| PR |
0.00 |
Add the number of warmup threads as a flag option. |
|
— |
2026-03-17 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Update Megascale debugging playbook. |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
[XLA:Runtime]: Remove unused method from buffer_sequencing_e |
|
— |
2026-03-19 |
| PR |
0.00 |
PR #39500: [ROCm] Add rocm_rbe_dynamic config |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-19 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Avoid logging that a test is being skipped repeatedly in eve |
|
— |
2026-03-18 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-13 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
| PR |
0.00 |
Automated Code Change |
|
— |
2026-03-20 |
