From 6aece5060b600f9c86580c5958b6039973a424fa Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 23 Nov 2016 18:06:54 +0800 Subject: [PATCH] Stash --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 11 +-- ...ence_nest_rnn_multi_unequalength_inputs.py | 69 ++++++++----------- 2 files changed, 36 insertions(+), 44 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index e1a847fc9ca..4d29507ca3c 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -139,20 +139,21 @@ 本例中的配置，使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 使用一个\ :code:`recurrent_group`\ 将两个序列同时过完全连接的\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ 的code如下。 -.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py :language: python - :lines: 41-58 + :lines: 42-59 :linenos: - 双层序列\： - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 - - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 + - 函数`outer_step`中可以分别处理这两个特征，但我们需要用\ :red:`targetInlink`\ 指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 -.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py :language: python - :lines: 41-89 + :lines: 42-75, 82-89 + :linenos: 示例4：beam_search的生成 ======================== diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py index 1b709a39c4b..bf88d00f2df 100644 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py @@ -1,4 +1,4 @@ -#edit-mode: -*- python -*- +# edit-mode: -*- python -*- # Copyright (c) 2016 Baidu, Inc. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); @@ -35,46 +35,37 @@ speaker2 = data_layer(name="word2", size=dict_dim) emb1 = embedding_layer(input=speaker1, size=word_dim) emb2 = embedding_layer(input=speaker2, size=word_dim) -# This hierachical RNN is designed to be equivalent to the simple RNN in -# sequence_rnn_multi_unequalength_inputs.conf - +# This hierarchical RNN is designed to be equivalent to the simple RNN in +# sequence_rnn_multi_unequalength_inputs.conf def outer_step(x1, x2): - outer_mem1 = memory(name="outer_rnn_state1", size=hidden_dim) - outer_mem2 = memory(name="outer_rnn_state2", size=hidden_dim) - - def inner_step1(y): - inner_mem = memory( - name='inner_rnn_state_' + y.name, - size=hidden_dim, - boot_layer=outer_mem1) - out = fc_layer( - input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name='inner_rnn_state_' + y.name) - return out - - def inner_step2(y): - inner_mem = memory( - name='inner_rnn_state_' + y.name, - size=hidden_dim, - boot_layer=outer_mem2) - out = fc_layer( - input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name='inner_rnn_state_' + y.name) - return out - - encoder1 = recurrent_group(step=inner_step1, name='inner1', input=x1) - - encoder2 = recurrent_group(step=inner_step2, name='inner2', input=x2) - - sentence_last_state1 = last_seq(input=encoder1, name='outer_rnn_state1') - sentence_last_state2_ = last_seq(input=encoder2, name='outer_rnn_state2') + index = [0] + + def inner_step(ipt): + index[0] += 1 + i = index[0] + outer_mem = memory(name="outer_rnn_state_%d" % i, size=hidden_dim) + + def inner_step_impl(y): + inner_mem = memory( + name="inner_rnn_state_" + y.name, + size=hidden_dim, + boot_layer=outer_mem) + out = fc_layer( + input=[y, inner_mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name='inner_rnn_state_' + y.name) + return out + + encoder = recurrent_group( + step=inner_step_impl, name='inner_%d' % i, input=ipt) + last = last_seq(name="outer_rnn_state_%d" % i, input=encoder) + return encoder, last + + _, sentence_last_state1 = inner_step(ipt=x1) + encoder2, _ = inner_step(ipt=x2) encoder1_expand = expand_layer( input=sentence_last_state1, expand_as=encoder2) -- GitLab