Commits

Branch selector

 master

User selector

deqiangc

All time

Commit History

Commits on Jun 7, 2024

• Try ResourceManager if a tensor is not found in IfrtRestoreTensorRegistry

  

  deqiangc

  authored and

  tensorflower-gardener

  committedJun 7, 2024
  a050cbf

Commits on Jun 6, 2024

• Run DecomposeResourceOpsPass before SinkVariableAsNamedArrayPass

  

  deqiangc

  authored and

  tensorflower-gardener

  committedJun 6, 2024
  91d5eca

Commits on Jun 5, 2024

• Add int4 support

  

  deqiangc

  authored and

  tensorflower-gardener

  committedJun 5, 2024
  97767a6

Commits on May 31, 2024

• Support cast in Ifrt restore operation

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 31, 2024
  6a8281f

• Add max_enqueued_batches option for model servers.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 31, 2024
  f34dc6a

• Avoid rewrite ReadVaraibleOps that are assigned in the same module

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 31, 2024
  bc97203

• Support DenseBoolArrayAttr in MLRT

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 31, 2024
  e0fc6ed

Commits on May 29, 2024

• Load variable concurrently

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 29, 2024
  4a4307e

Commits on May 24, 2024

• Allow specifying compilation environment for ifrt compilation

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 24, 2024
  f471831

Commits on May 22, 2024

• Remove unused attributes from IfrtLoadVariableOp

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 22, 2024
  fc2f085

Commits on May 17, 2024

• Make returned future immdiately ready if it is TPU used only.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 17, 2024
  45fa10c

Commits on May 16, 2024

• Add a new test case in pjrt_util_test

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 16, 2024
  3c92d1e

Commits on May 15, 2024

• Add unit test for async_value_tensor.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 15, 2024
  1327313

Commits on May 14, 2024

• tf_host_callback in tfrt/ifrt use DeviceMgr instead of StaticDeviceMgr

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 14, 2024
  2fca883

• Reduce messages from sharding_utils.cc

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 14, 2024
  e3e649b

Commits on May 8, 2024

• Fix asan error due to additional error logging

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 8, 2024
  eb93fb5

• Add additional logging when unwinding error in mlrt

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 8, 2024
  5a9da02

Commits on May 6, 2024

• Implement Freeze() in IfrtServingExecutable to release mlir module

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 6, 2024
  7b26bb0

• Add integration options in TFRT/IFRT with tensorflow serving.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 6, 2024
  fedcb89

Commits on May 1, 2024

• Implement a Freeze() api to release host memory for device tensors.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMay 1, 2024
  d9ae227

Commits on Apr 26, 2024

• Register and load all applicable dialects for parallel compilation

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 26, 2024
  dcd302e

Commits on Apr 25, 2024

• Make CachedExecutableBundle movable only to avoid repeated copy and

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 25, 2024
  612cea9

Commits on Apr 24, 2024

• Remove redundant load of QuantizationDialect

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 24, 2024
  3f24ddb

• Allow XlaHostComputeOp to the allowed list in mark_ops_for_outside_compilation.cc

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 24, 2024
  dadbf6d

Commits on Apr 18, 2024

• Provide a runtime option to lower bound the number of batch threads.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 18, 2024
  45f30ac

Commits on Apr 13, 2024

• IfrtLoadVariable current returns a key to look for loaded variable.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 13, 2024
  9b2e656

Commits on Apr 11, 2024

• Inject TpuCompiler pass to IfrtBackendCompiler

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 11, 2024
  1c56b45

Commits on Apr 10, 2024

• Clean up ownership of IfrtServingExecutable.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 10, 2024
  15a44c7

Commits on Apr 6, 2024

• IfrtServingExecutable support host callback execution

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 6, 2024
  cccd18a

Commits on Apr 4, 2024

• Create an executable wrapper to run host callback.

  

  deqiangc

  authored and

  tensorflower-gardener

  committedApr 4, 2024
  d39012e

Commits on Mar 26, 2024

• Change tf.IfrtLoadVariableOp to receive tensor future from

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMar 26, 2024
  4c65498

Commits on Mar 25, 2024

• Add a pass to remove unused restore op

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMar 25, 2024
  5e24f96

Commits on Mar 23, 2024

• Open source IdentityPropogation pass

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMar 23, 2024
  038b5c3

Commits on Mar 20, 2024

• Fix a use after free error in LowerToIfrtRestoreVariable pass

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMar 20, 2024
  c7c3bbd

Commits on Mar 19, 2024

• Lower tf.IfrtRestoreVariableOp to tf_mlrt.IfrtRestoreVariableOp

  

  deqiangc

  authored and

  tensorflower-gardener

  committedMar 19, 2024
  ac3c580

Pagination

PreviousNext