Commits
Branch selector
User selector
Commit History
Commits on Jan 15, 2022
Add MLIR implementation and unify tf.XlaSelectAndScatter op
Show description for aae314ccommitted[XLA] QoL improvement for --xla_dump_fusion_visualization: visualize the producer we are about to fuse
Show description for 5dc1045authored andcommitted[tf.data] Add a 0% experiment that sets the initial `parallelism` value for autotuned parallelizable Ops to a small value.
Show description for 7ab271bauthored andcommitted[MLIR][MHLO] Add a pass to collapse reductions to 1D or 2D equivalents
Show description for 3ad9798authored andcommitted
Introduce the TypeDispatchTable Class
Show description for d2f89f4authored andcommitted[XLA] QoL improvement for --xla_dump_fusion_visualization: try to higlight (unfused) producer
Show description for c319717authored andcommittedMerge pull request #53765 from perfinion:systemlib-abseil-cpp
Show description for 316bbe8committed[XLA] Correction to rendering of --xla_gpu_fusion_visualization: render after dead instruction is removed
Show description for 4df69faauthored andcommittedMerge pull request #53773 from tensorflow:mihaimaruseac-set-up-scorecards
Show description for dfbea8ecommittedSet up an XLA/TPU regression test that compiles+executes a given SavedModel.
Show description for 5ad1157committed- authored
Commits on Jan 14, 2022
[tf:jitrt] Rename all occurrences of cpurt to jitrt
Show description for 0385773authored andcommittedAlways generate a context_id in TraceMeProducer
Show description for 61154f2authored andcommittedAdd adagrad + momentum optimizer in tf2
Show description for 4865433committed[tf.data service] Static sharding does not require local reading.
Show description for 829946eauthored andcommittedRemove the option to use the cudnn variant of batch normalization in XLA:GPU.
Show description for 450b48bauthored andcommittedMove out FunctionSpec from function.py
Show description for a7b687fauthored andcommitted[tf.data] Simplify how transformation names are set.
Show description for 441a372authored andcommittedRefactor HostTracer and PythonTracer libraries
Show description for f315091authored andcommitted[XLA] Rendering fixes for --xla_dump_fusion_visualization
Show description for 256ac0eauthored andcommittedMigrate usages of deprecated Trackable methods to `_trackable_children`.
Show description for 6b1287dauthored andcommittedAdd TpuEmbedding Enqueue API's.
Show description for 6bfa530authored andcommittedAdding setter for use_global_device_ids in HloAllReduceInstructionBase.
Show description for f21fffeauthored andcommittedEnsure we track fused batchnorm layer's trainable variables when the batch is empty to avoid an edge case with distributed training.
Show description for 680551bcommittedAdd checkpoint size instrumentation to Saver checkpoint API.
Show description for 2a3b691authored andcommittedChange LOG(INFO) to VLOG(1) to give more fine grained control of logging.
Show description for e4b62daauthored andcommittedRemove `std=c++14` in `BUILD` files.
Show description for bfea11aauthored andcommittedCreate an accessor for the EagerContext object owned by the "current eager Python context object". Because so many things are tightly coupled to that Python context object, and that currently compl…
Show description for 8ed106fauthored andcommittedInternal visibility change
Show description for d13e5f9authored andcommittedDisable test that times out on GPU.
Show description for a6a60c5authored andcommittedRemove stale comment and fix pre C++14 code.
Show description for b7b15afauthored andcommitted[tf:jitrt] Rename all tf_cpurt files to tf_jitrt
Show description for 05fff77authored andcommitted- committed
Move forward compatibility date of anonymous_iterator_v3 out by a week.
Show description for b9cb013authored andcommittedUpdate TFRT dependency to use revision
Show description for 949c3d4committed