1395fb42 · [PJRT:GPU] Don't use XLA transfer manager APIs to perform D2H copies. · 4月 11, 2024

3b2b117a · (1/N) Remove StableHLO Quantizer quantize pass dependency to preset. · 4月 12, 2024

8898f77e · Automated Code Change · 4月 12, 2024

4bf7242f · Transpose weights for hybrid quantized convolution · 4月 12, 2024

239223f1 · Pass a flag whether we are in code emission phase. · 4月 12, 2024

c09fa8eb · Transpose weights for hybrid quantized convolution · 4月 12, 2024

028f35d6 · [XLA] Fix issue in MHLO->HLO where a constant with less than 32 elements... · 4月 12, 2024

e520fb38 · When no weight cache is provided to XNNPack, create one to share packed weights between operations. · 4月 12, 2024

dd9a8d4a · Update the XNNPack version. · 4月 12, 2024

aaf512d4 · [XLA:SPMD] Add DUS to aggressiveness 0. · 4月 12, 2024

590b8a7a · Define MtkNeuronSettings in configuration.proto · 4月 12, 2024

b2821a00 · [xla][gpu] Fixed two problems in p2p-schedule-preparation. · 4月 12, 2024

3a354648 · Integrate StableHLO at openxla/stablehlo@714d9aca · 4月 12, 2024

a9614ea3 · Add unbounded dynamism test for ReduceScatterOp. · 4月 12, 2024

05e106d3 · Optimize `BatchedDevicePut` by batching GIL release · 4月 12, 2024

f3b27c34 · Add prettier printing for constant operations to make it easier to debug... · 4月 12, 2024

0621056b · PR #11299: [XLA:CPU] Enable XLA on Windows · 4月 12, 2024

b8511d13 · [XLA] Add support for two-bit types · 4月 12, 2024

a3ec5a0f · Move #include <stddef.h> out of namespace. · 4月 12, 2024

d4305131 · [XLA] Clear derived instruction's sharding only if shapes are incompatible. · 4月 12, 2024