You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2021/10/17 00:44:56 UTC
[GitHub] [tvm] apivovarov opened a new pull request #9303: [Keras] Support return_sequences in LSTM
apivovarov opened a new pull request #9303:
URL: https://github.com/apache/tvm/pull/9303
This PR adds support for `return_sequences` bool param for Keras LSTM layer.
By default the param is False - LSTM returns only the last h output
If `return_sequences is True` then ALL h outputs should be returned. The outputs should be concatenated along axis 1.
I also added dedicated test for LSTM. Input shape increased to test `return_sequences is True` case.
[Understanding LSTM Networks](https://colah.github.io/posts/2015-08-Understanding-LSTMs/)
Related discussion https://discuss.tvm.apache.org/t/keras-lstm-return-sequences-parameter-not-supported/11239
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] vinx13 merged pull request #9303: [Keras] Support return_sequences in LSTM
Posted by GitBox <gi...@apache.org>.
vinx13 merged pull request #9303:
URL: https://github.com/apache/tvm/pull/9303
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #9303: [Keras] Support return_sequences in LSTM
Posted by GitBox <gi...@apache.org>.
AndrewZhaoLuo commented on a change in pull request #9303:
URL: https://github.com/apache/tvm/pull/9303#discussion_r731369633
##########
File path: tests/python/frontend/keras/test_forward.py
##########
@@ -417,12 +417,20 @@ def test_forward_reuse_layers(self, keras):
keras_model = keras.models.Model(data, z)
verify_keras_frontend(keras_model)
+ def test_forward_lstm(self, keras):
+ data = keras.layers.Input(shape=(10, 32))
+ rnn_funcs = [
+ keras.layers.LSTM(16),
+ keras.layers.LSTM(16, return_sequences=True),
+ ]
+ for rnn_func in rnn_funcs:
+ x = rnn_func(data)
+ keras_model = keras.models.Model(data, x)
+ verify_keras_frontend(keras_model, need_transpose=False)
+
def test_forward_rnn(self, keras):
data = keras.layers.Input(shape=(1, 32))
rnn_funcs = [
- keras.layers.LSTM(
Review comment:
i would keep this test here for completeness and make the above only testing the return_sequence flag.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] apivovarov commented on pull request #9303: [Keras] Support return_sequences in LSTM
Posted by GitBox <gi...@apache.org>.
apivovarov commented on pull request #9303:
URL: https://github.com/apache/tvm/pull/9303#issuecomment-946133766
@Huyuwei @mbrookhart @kazum @yongwww Guys, can you help to review this PR?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] apivovarov commented on a change in pull request #9303: [Keras] Support return_sequences in LSTM
Posted by GitBox <gi...@apache.org>.
apivovarov commented on a change in pull request #9303:
URL: https://github.com/apache/tvm/pull/9303#discussion_r731380430
##########
File path: tests/python/frontend/keras/test_forward.py
##########
@@ -417,12 +417,20 @@ def test_forward_reuse_layers(self, keras):
keras_model = keras.models.Model(data, z)
verify_keras_frontend(keras_model)
+ def test_forward_lstm(self, keras):
+ data = keras.layers.Input(shape=(10, 32))
+ rnn_funcs = [
+ keras.layers.LSTM(16),
+ keras.layers.LSTM(16, return_sequences=True),
+ ]
+ for rnn_func in rnn_funcs:
+ x = rnn_func(data)
+ keras_model = keras.models.Model(data, x)
+ verify_keras_frontend(keras_model, need_transpose=False)
+
def test_forward_rnn(self, keras):
data = keras.layers.Input(shape=(1, 32))
rnn_funcs = [
- keras.layers.LSTM(
Review comment:
ok, I returned LSTM test back to test_forward_rnn. Lets also test both LSTMs (with and without return_sequence flag) in `test_forward_lstm`. New test `test_forward_lstm` uses more realistic input with shape `(10, 32)` instead of `(1, 32)`. Using one word for LSTM input does not test the Recursion logic inside LSTM impl.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org