| Commit message (Collapse) | Author | Age |
|
|
|
| |
Change: 142805270
|
|
|
|
|
| |
SavedModel export formats.
Change: 142803996
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change modifies the GrpcWorkerCache so that, when a master
attempts to communicate with the worker in the same process, it does
so by direct method calls on a `WorkerInterface*`, without making a
loopback RPC call.
This change is another incremental step towards addressing issue
#6256. There are further improvements possible, and we will continue
to investigate them, including:
* Avoiding the protobuf encoding/decoding for request/response objects
where this affects performance. The zero-copy `TensorResponse` class
is an example of how we could improve performance here, for
`RunGraphRequest` and `RunGraphResponse` objects.
* Profiling the closure creation/context switch overhead for
interactions with the local worker.
Change: 142793965
|
|
|
|
| |
Change: 142791193
|
|
|
|
| |
Change: 142790604
|
|
|
|
|
|
| |
more imports to this file.
Change: 142784743
|
|
|
|
| |
Change: 142783149
|
|
|
|
| |
Change: 142781367
|
|
|
|
|
|
|
|
|
| |
This makes print-out messages less random, more informative, and shorter in most cases. For example:
Before: <third_party.tensorflow.python.framework.ops.Operation object at 0x7f6e003c6f50>
After: <tf.Operation 'train_op/dnn/control_dependency' type=Identity>
It also improves consistency with Tensor.__repr__().
Change: 142781125
|
|
|
|
|
|
| |
duplication.
Change: 142780852
|
|
|
|
| |
Change: 142779313
|
|
|
|
| |
Change: 142778020
|
|
|
|
| |
Change: 142777129
|
|
|
|
| |
Change: 142775917
|
|
|
|
| |
Change: 142767321
|
|
|
|
|
|
| |
latest gemmlowp, which makes it an error by default to build without
SSE4.1 on x86 or NEON on ARM.
Change: 142758203
|
|
|
|
| |
Change: 142748584
|
|
|
|
| |
Change: 142732418
|
|
|
|
| |
Change: 142730769
|
|
|
|
| |
Change: 142729497
|
|
|
|
| |
Change: 142728126
|
|
|
|
| |
Change: 142727588
|
|
|
|
| |
Change: 142726205
|
|
|
|
| |
Change: 142724724
|
|
|
|
|
|
| |
with infer_real_valued_columns_from_input_fn and canned estimators.
Change: 142719058
|
|
|
|
|
| |
of Define() where that doesn't work, switch to Create() instead.
Change: 142718606
|
|
|
|
| |
Change: 142718232
|
|
|
|
| |
Change: 142717732
|
|
|
|
| |
Change: 142716019
|
|
|
|
|
| |
shape parameter.
Change: 142713914
|
|
|
|
| |
Change: 142707321
|
|
|
|
| |
Change: 142705711
|
|
|
|
| |
Change: 142700768
|
|
|
|
| |
Change: 142698388
|
|
|
|
|
|
| |
explaining how to load from assets, and also log warning if compressed proto is larger than 64mb.
Change: 142698084
|
|
|
|
| |
Change: 142694447
|
|
|
|
| |
Change: 142686873
|
|
|
|
| |
Change: 142686791
|
|
|
|
|
|
| |
upper or lower bound.
Change: 142684404
|
|
|
|
|
| |
Empirically, n ~ 75 gives the most speedup for the bootstrap method.
Change: 142684023
|
|
|
|
| |
Change: 142683937
|
|
|
|
|
|
| |
summary API convention
Change: 142682974
|
|
|
|
| |
Change: 142682138
|
|
|
|
|
| |
This brings the codebase in closer alignment with the style guide: "... and do not abbreviate by deleting letters within a word."
Change: 142681857
|
|
|
|
|
| |
This changelist adds the necessary C++ and Python code to register a new op to partition and read BigQuery tables.
Change: 142681535
|
|
|
|
|
| |
multiplication. Needs a new enough version of libxsmm with sparse support, and needs some patches to work reliably at all sizes.
Change: 142680668
|
|
|
|
| |
Change: 142677515
|
|
|
|
| |
Change: 142676422
|
|
|
|
| |
Change: 142673866
|
|
|
|
| |
Change: 142669605
|