[TF]Refine LSTMBlockCell to support dynamic rnn #5963
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
lixiaoquan
changed the title
lixiaoquan
force-pushed
the
lstm
branch
3 times, most recently
from
7460648 to
9acbe2c
Compare
yongwww
requested changes
Jul 8, 2020
| # Return dummy for those unused values | ||
| dummy = tvm.relay.const(0) | ||
| return tvm.relay.TupleWrapper( | ||
| tvm.relay.Tuple([dummy, next_c, dummy, dummy, dummy, dummy, next_h]), 7) |
There was a problem hiding this comment.
I am not sure if the dummy node will be used somehow in some cases. It would be good to generate the real node as TF does for the dummy nodes.
There was a problem hiding this comment.
def lstm_block_cell(x, cs_prev, h_prev, w, wci, wcf, wco, b, forget_bias=1, cell_clip=3, use_peephole=False, name=None):
r"""Computes the LSTM cell forward propagation for 1 time step.
This implementation uses 1 weight matrix and 1 bias vector, and there's an
optional peephole connection.
This kernel op implements the following mathematical equations:
xh = [x, h_prev]
[i, f, ci, o] = xh * w + b
f = f + forget_bias
if not use_peephole:
wci = wcf = wco = 0
i = sigmoid(cs_prev * wci + i)
f = sigmoid(cs_prev * wcf + f)
ci = tanh(ci)
cs = ci .* i + cs_prev .* f
cs = clip(cs, cell_clip)
o = sigmoid(cs * wco + o)
co = tanh(cs)
h = co .* o
...
Returns:
A tuple of `Tensor` objects (i, cs, f, o, ci, co, h).
...
There was a problem hiding this comment.
I added all return objects, but I just can't test them, because tensorflow.contrib.rnn.LSTMBlockCell.call() only return h and new_states
I think if users uses API to contruct graph, other values won't be used.
(cs_prev, h_prev) = state
(_, cs, _, _, _, _, h) = _lstm_block_cell(
inputs,
cs_prev,
h_prev,
self._kernel,
self._bias,
wci=wci,
wcf=wcf,
wco=wco,
forget_bias=self._forget_bias,
cell_clip=self._cell_clip,
use_peephole=self._use_peephole)
new_state = rnn_cell_impl.LSTMStateTuple(cs, h)
return h, new_state
There was a problem hiding this comment.
cc @yongwww Could you take another look? Thanks
1. Refine conversion of `LSTMBlockCell`
1) Make its output follows definition in TensorFlow
2) Avoid introducing variables which doesn't match any placeholder nodes in TensorFlow graph
2. About change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by convert those states node into placeholders.
|
ping @zhiics Could you help to review and merge？ Thanks |
|
Thanks @lixiaoquan @yongwww |
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Aug 26, 2020
1. Refine conversion of `LSTMBlockCell`
1) Make its output follows definition in TensorFlow
2) Avoid introducing variables which doesn't match any placeholder nodes in TensorFlow graph
2. About change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by convert those states node into placeholders.
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Aug 26, 2020
1. Refine conversion of `LSTMBlockCell`
1) Make its output follows definition in TensorFlow
2) Avoid introducing variables which doesn't match any placeholder nodes in TensorFlow graph
2. About change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by convert those states node into placeholders.
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Sep 2, 2020
1. Refine conversion of `LSTMBlockCell`
1) Make its output follows definition in TensorFlow
2) Avoid introducing variables which doesn't match any placeholder nodes in TensorFlow graph
2. About change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by convert those states node into placeholders.
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Sep 3, 2020
1. Refine conversion of `LSTMBlockCell`
1) Make its output follows definition in TensorFlow
2) Avoid introducing variables which doesn't match any placeholder nodes in TensorFlow graph
2. About change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by convert those states node into placeholders.
|
Tensorflow2 LSTM failed - the discussion is here https://discuss.tvm.apache.org/t/tensorflow2-lstm-failed/11174 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Refine conversion of
LSTMBlockCellAbout change in test_forward_ptb
States nodes of LSTMBlockCell in this PB file are actually Constant node.
TF can feed data to those Constant nodes but relay can't do that, so current conversion of LSTMBockCell introduces extra variables to solve this issue.
But this causes that relay IR doesn't match original TF graph. This PR solves this issue by converting those states node into placeholders.
cc @kevinthesun @srkreddy1238 @joyalbin Please help to review, thanks