experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Fix typo.

Browse Source 
PiperOrigin-RevId: 238456026
This commit is contained in:
Shashi Shekhar 2019-03-14 09:19:15 -07:00 committed by TensorFlower Gardener
parent 872b9bd9fc
commit 61ebe6e8f6
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/lite/experimental/examples/lstm/g3doc/README.md
View File

@ -101,7 +101,7 @@ tflite_model = converter.convert() # You got a tflite model!
## Why introduce another set of LSTM APIs? ## Why introduce another set of LSTM APIs?
Bridging TensorFlow LSTM and TensorFlow Lite is not easy, and the use of Bridging TensorFlow LSTM and TensorFlow Lite is not easy, and the use of
`dynamic_rnn` adds addtional complexity (as the while loop is introduced). `dynamic_rnn` adds additional complexity (as the while loop is introduced).
With the help of With the help of
[OpHint](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/python/op_hint.py) [OpHint](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/python/op_hint.py)
(also see the next section), we create special wrappers around `rnn_cell` and (also see the next section), we create special wrappers around `rnn_cell` and
@ -373,7 +373,7 @@ See below.
`tf.lite.experimental.nn.bidirectional_dynamic_rnn` only supports `tf.lite.experimental.nn.bidirectional_dynamic_rnn` only supports
`control_flow_v2`, you can this on by setting the environment variable `control_flow_v2`, you can this on by setting the environment variable
`TF_ENABLE_CONTROL_FLOW_V2=1`, see in the tutorial. `TF_ENABLE_CONTROL_FLOW_V2=1`, see in the tutorial.
* Currently, `sequence_length` is not supported, perfer to set it to None. * Currently, `sequence_length` is not supported, prefer to set it to None.
* `num_unit_shards` & `num_proj_shards` in LSTMCell are not supported as * `num_unit_shards` & `num_proj_shards` in LSTMCell are not supported as
well. well.
* Currently, `tf.lite.experimental.nn.dynamic_rnn` & * Currently, `tf.lite.experimental.nn.dynamic_rnn` &