Commits

Branch selector

 master

User selector

protoget

Datepicker

All time

Commit History

Commits on Nov 14, 2017

• Delete infeasible TODO in CudnnRNN

  Show description for 579276a
  

  protoget

  authored and

  tensorflower-gardener

  committed
  579276a

  Copy full SHA for 579276a

Commits on Nov 13, 2017

• Fix CuDNNCompatibleGRU after GRUCell refactorization

  Show description for 90222dd
  

  protoget

  authored and

  tensorflower-gardener

  committed
  90222dd

  Copy full SHA for 90222dd

• Fix build breakage

  Show description for 4d5e805
  

  protoget

  authored and

  tensorflower-gardener

  committed
  4d5e805

  Copy full SHA for 4d5e805

Commits on Nov 11, 2017

• Include memory size info in error message.

  Show description for 8d15389
  

  protoget

  authored and

  aselle

  committed
  8d15389

  Copy full SHA for 8d15389

Commits on Nov 3, 2017

• Switch tf.contrib.cudnn_rnn.CudnnXXX to point to layer APIs instead of op wrappers

  Show description for 823d8d4
  

  protoget

  authored and

  tensorflower-gardener

  committed
  823d8d4

  Copy full SHA for 823d8d4

• Support Cudnn RNN Fp16

  Show description for 509d0f2
  

  protoget

  authored and

  tensorflower-gardener

  committed
  509d0f2

  Copy full SHA for 509d0f2

Commits on Oct 26, 2017

• Switch tf.contrib.cudnn_rnn.CudnnXXX to point to layer APIs instead of op wrappers

  Show description for 62f62bf
  

  protoget

  authored and

  tensorflower-gardener

  committed
  62f62bf

  Copy full SHA for 62f62bf

Commits on Oct 6, 2017

• Fix a minor issue w/ allreduce

  Show description for af6e00f
  

  protoget

  authored and

  tensorflower-gardener

  committed
  af6e00f

  Copy full SHA for af6e00f

• Fix a minor issue w/ allreduce

  Show description for 71a2859
  

  protoget

  authored and

  tensorflower-gardener

  committed
  71a2859

  Copy full SHA for 71a2859

Commits on Oct 5, 2017

• Introduce CudnnRNN layers

  Show description for 2ae5bfc
  

  protoget

  authored and

  tensorflower-gardener

  committed
  2ae5bfc

  Copy full SHA for 2ae5bfc

Commits on Sep 28, 2017

• [Adagrad optimizer] Add support for dynamic shape variable

  Show description for 163f42e
  

  protoget

  authored and

  tensorflower-gardener

  committed
  163f42e

  Copy full SHA for 163f42e

Commits on Aug 31, 2017

• Add functional cudnn_rnn_ops

  Show description for 4807158
  

  protoget

  authored and

  tensorflower-gardener

  committed
  4807158

  Copy full SHA for 4807158

Commits on Aug 30, 2017

• Exposing CudnnCompatibleRNN classes.

  Show description for 8f1746c
  

  protoget

  authored and

  tensorflower-gardener

  committed
  8f1746c

  Copy full SHA for 8f1746c

Commits on Aug 24, 2017

• Revamp CudnnRNN Saveables

  Show description for b30ce47
  

  protoget

  authored and

  tensorflower-gardener

  committed
  b30ce47

  Copy full SHA for b30ce47

Commits on Aug 22, 2017

• Change CudnnCompatibleLSTMCell to wrap LSTMBlockCell

  Show description for d4a5589
  

  protoget

  authored and

  tensorflower-gardener

  committed
  d4a5589

  Copy full SHA for d4a5589

• Delete a few cudnn rnn unittests

  Show description for 675a9dc
  

  protoget

  authored and

  tensorflower-gardener

  committed
  675a9dc

  Copy full SHA for 675a9dc

Commits on Aug 16, 2017

• Return correct dynamic shape from CuDNN RNN ParamsToCanonical kernel

  Show description for f39ed67
  

  protoget

  authored and

  tensorflower-gardener

  committed
  f39ed67

  Copy full SHA for f39ed67

Commits on Aug 9, 2017

• slot_creator: fix bugs handling dynamic-shaped var/tensor

  Show description for e6e7ee4
  

  protoget

  authored and

  tensorflower-gardener

  committed
  e6e7ee4

  Copy full SHA for e6e7ee4

Commits on Aug 2, 2017

• Support bidi-Cudnn LSTM save/restore.

  Show description for 320597a
  

  protoget

  authored and

  tensorflower-gardener

  committed
  320597a

  Copy full SHA for 320597a

Commits on Jul 18, 2017

• Support float64 CuDNN RNN

  Show description for ed03ee3
  

  protoget

  authored and

  tensorflower-gardener

  committed
  ed03ee3

  Copy full SHA for ed03ee3

Commits on Jul 4, 2017

• (unidirectional)Cudnn compatible GRU Cell

  Show description for 03f44a7
  

  protoget

  authored and

  tensorflower-gardener

  committed
  03f44a7

  Copy full SHA for 03f44a7

Commits on Jun 27, 2017

• Cudnn compatible LSTMCell and LSTMBlockCell

  Show description for 2b5d00a
  

  protoget

  authored and

  tensorflower-gardener

  committed
  2b5d00a

  Copy full SHA for 2b5d00a

Commits on Jun 20, 2017

• Support reuse cuDNNLSTM-trained checkpoints by multi-layer LSTM(Block)Cell

  Show description for a936b23
  

  protoget

  authored and

  tensorflower-gardener

  committed
  a936b23

  Copy full SHA for a936b23

Commits on Jun 1, 2017

• Fix graph text format serialization

  Show description for 2462365
  

  protoget

  authored and

  tensorflower-gardener

  committed
  2462365

  Copy full SHA for 2462365

Commits on May 22, 2017

• Use Cudnn for avgpool by default.

  Show description for 9cc6f65
  

  protoget

  authored and

  tensorflower-gardener

  committed
  9cc6f65

  Copy full SHA for 9cc6f65

Commits on May 15, 2017

• Change CudnnRNN default input mode from "auto_select" to "linear_input".

  Show description for 8831aa7
  

  protoget

  authored and

  tensorflower-gardener

  committed
  8831aa7

  Copy full SHA for 8831aa7

Commits on May 6, 2017

• Support dropout in cudnn RNN.

  Show description for 6c02261
  

  protoget

  authored and

  tensorflower-gardener

  committed
  6c02261

  Copy full SHA for 6c02261

Commits on Apr 26, 2017

• Fix two tiny bugs in cudnn_rnn_ops

  Show description for 5c2c39d
  

  protoget

  authored and

  tensorflower-gardener

  committed
  5c2c39d

  Copy full SHA for 5c2c39d

Commits on Mar 9, 2017

• Add sparsify_gather op to reduce linear model memory footprint.

  Show description for 168e8da
  

  protoget

  authored and

  tensorflower-gardener

  committed
  168e8da

  Copy full SHA for 168e8da

Commits on Mar 1, 2017

• Fix TF-learn eval_evalset/eval_evaltrain job failure due to race condition.

  Show description for 9560e41
  

  protoget

  authored and

  tensorflower-gardener

  committed
  9560e41

  Copy full SHA for 9560e41