45acdddf · Change tf.IfrtLoadVariableOp to receive tensor future from · 3月 20, 2024

3d2ff8f2 · Apply the sharding on RngBitGeneratorOp itself, instead of its tuple elements. · 3月 20, 2024

37b15ff9 · Fix JAX PjRT Clients using XLA pins from 12/2023. · 3月 20, 2024

c54a8737 · Change flat_hash_map to vector to have deterministic order while unrolling. · 3月 20, 2024

5467f56f · Add `gpu_h100` backend to the `tf_xla_py_test` build macro. · 3月 20, 2024

9887148a · Supports PjRt CPU array converting py array when the CPU PjRt arrays have non-default layouts. · 3月 20, 2024

cacbed03 · Add unbounded dynamism test for for Atan2Op. · 3月 20, 2024

5ea691e0 · Add missing includes and fix typo. · 3月 20, 2024

9b767639 · Add unbounded dynamism test for BitcastConvertOp. · 3月 20, 2024

1d294381 · Print out a warning if an instruction with misaligned sharding annotations. · 3月 20, 2024

ebf05ad8 · Use the appropriate operand sharding when computing resharding costs for HLO... · 3月 20, 2024

a9e8212a · Changes visibility for metrics_proto to public. · 3月 20, 2024

1e1d3d50 · [XLA:GPU] Add a Gemv rewriter pass to convert gemv to gemm with a trivial dimension · 3月 20, 2024

7a358eae · Migrate `UniformSupport*` and `FakeQuantSupport*` components from `lite` directory. · 3月 20, 2024

54c0a5e5 · Ignore nested tuples, if present, during post processing. · 3月 20, 2024

a0a87abc · Move `tsl/util` to `xla/tsl/util` · 3月 20, 2024

9297994a · Migrate direct callers of options().packed() to is_packed() helper. · 3月 20, 2024