# SOME DESCRIPTIVE TITLE. # Copyright (C) 2021, PaddleNLP # This file is distributed under the same license as the PaddleNLP package. # FIRST AUTHOR , 2022. # #, fuzzy msgid "" msgstr "" "Project-Id-Version: PaddleNLP \n" "Report-Msgid-Bugs-To: \n" "POT-Creation-Date: 2022-03-18 21:31+0800\n" "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n" "Last-Translator: FULL NAME \n" "Language-Team: LANGUAGE \n" "MIME-Version: 1.0\n" "Content-Type: text/plain; charset=utf-8\n" "Content-Transfer-Encoding: 8bit\n" "Generated-By: Babel 2.9.0\n" #: ../source/paddlenlp.layers.crf.rst:2 msgid "crf" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf:1 msgid "" "LinearChainCrf is a linear chain Conditional Random Field layer, it can " "implement sequential dependencies in the predictions. Therefore, it can " "take context into account whereas a classifier predicts a label for a " "single sample without considering \"neighboring\" samples. See " "https://repository.upenn.edu/cgi/viewcontent.cgi?article=1162&context=cis_papers" " for reference." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf #: paddlenlp.layers.crf.LinearChainCrf.forward #: paddlenlp.layers.crf.LinearChainCrf.gold_score #: paddlenlp.layers.crf.LinearChainCrfLoss #: paddlenlp.layers.crf.LinearChainCrfLoss.forward #: paddlenlp.layers.crf.ViterbiDecoder #: paddlenlp.layers.crf.ViterbiDecoder.forward msgid "参数" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf:5 msgid "The label number." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf:7 msgid "The crf layer learning rate. Defaults to ``0.1``." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf:9 msgid "" "If set to True, the start tag and stop tag will be considered, the " "transitions params will be a tensor with a shape of `[num_labels+2, " "num_labels+2]`. Else, the transitions params will be a tensor with a " "shape of `[num_labels, num_labels]`." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:1 msgid "" "Computes the normalization in a linear-chain CRF. See " "http://www.cs.columbia.edu/~mcollins/fb.pdf for reference." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:3 msgid "" "F & = logZ(x) = log\\sum_y exp(score(x,y))\n" "\n" "score(x,y) & = \\sum_i Emit(x_i,y_i) + Trans(y_{i-1}, y_i)\n" "\n" "p(y_i) & = Emit(x_i,y_i), T(y_{i-1}, y_i) = Trans(y_{i-1}, y_i)" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:10 msgid "then we can get:" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:12 msgid "" "F(1) = log\\sum_{y1} exp(p(y_1) + T([START], y1))\n" "\n" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:15 msgid "" "F(2) & = log\\sum_{y1}\\sum_{y2} exp(p(y_1) + T([START], y1) + p(y_2) + " "T(y_1,y_2)) \\\\\n" "& = log\\sum_{y2} exp(F(1) + p(y_2) + T(y_1,y_2))\n" "\n" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:19 msgid "Further, We can get F(n) is a recursive formula with F(n-1)." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:21 #: paddlenlp.layers.crf.LinearChainCrf.gold_score:4 #: paddlenlp.layers.crf.LinearChainCrfLoss.forward:4 msgid "" "The input predicted tensor. Its dtype is float32 and has a shape of " "`[batch_size, sequence_length, num_tags]`." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:23 #: paddlenlp.layers.crf.LinearChainCrf.gold_score:8 #: paddlenlp.layers.crf.LinearChainCrfLoss.forward:6 msgid "The input length. Its dtype is int64 and has a shape of `[batch_size]`." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward #: paddlenlp.layers.crf.LinearChainCrf.gold_score #: paddlenlp.layers.crf.LinearChainCrfLoss.forward #: paddlenlp.layers.crf.ViterbiDecoder.forward msgid "返回" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward:26 msgid "" "Returns the normalizers tensor `norm_score`. Its dtype is float32 and has" " a shape of `[batch_size]`." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.forward #: paddlenlp.layers.crf.LinearChainCrf.gold_score #: paddlenlp.layers.crf.LinearChainCrfLoss.forward #: paddlenlp.layers.crf.ViterbiDecoder.forward msgid "返回类型" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.gold_score:1 msgid "" "Computes the unnormalized score for a tag sequence. $$ score(x,y) = " "\\sum_i Emit(x_i,y_i) + Trans(y_{i-1}, y_i) $$" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.gold_score:6 #: paddlenlp.layers.crf.LinearChainCrfLoss.forward:8 msgid "" "The input label tensor. Its dtype is int64 and has a shape of " "`[batch_size, sequence_length]`" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrf.gold_score:11 msgid "" "Returns the unnormalized sequence scores tensor `unnorm_score`. Its dtype" " is float32 and has a shape of `[batch_size]`." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrfLoss:1 msgid "" "The negative log-likelihood for linear chain Conditional Random Field " "(CRF)." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrfLoss:3 msgid "" "The `LinearChainCrf` network object. Its parameter will be used to " "calculate the loss." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrfLoss.forward:1 msgid "" "Calculate the crf loss. Let $$ Z(x) = \\sum_{y'}exp(score(x,y')) $$, " "means the sum of all path scores, then we have $$ loss = -logp(y|x) = " "-log(exp(score(x,y))/Z(x)) = -score(x,y) + logZ(x) $$" msgstr "" #: of paddlenlp.layers.crf.LinearChainCrfLoss.forward:10 msgid "" "Unnecessary parameter for compatibility with older versions. Defaults to " "``None``." msgstr "" #: of paddlenlp.layers.crf.LinearChainCrfLoss.forward:13 msgid "The crf loss. Its dtype is float32 and has a shape of `[batch_size]`." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder:1 msgid "" "ViterbiDecoder can decode the highest scoring sequence of tags, it should" " only be used at test time." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder:3 msgid "" "The transition matrix. Its dtype is float32 and has a shape of " "`[num_tags, num_tags]`." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder:5 msgid "" "If set to True, the last row and the last column of transitions will be " "considered as start tag, the penultimate row and the penultimate " "column of transitions will be considered as stop tag. Else, all the rows " "and columns will be considered as the real tag. Defaults to ``None``." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder.forward:1 msgid "Decode the highest scoring sequence of tags." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder.forward:3 msgid "" "The unary emission tensor. Its dtype is float32 and has a shape of " "`[batch_size, sequence_length, num_tags]`." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder.forward:5 msgid "" "The input length tensor storing real length of each sequence for " "correctness. Its dtype is int64 and has a shape of `[batch_size]`." msgstr "" #: of paddlenlp.layers.crf.ViterbiDecoder.forward:8 msgid "" "Returns tuple (scores, paths). The `scores` tensor containing the score " "for the Viterbi sequence. Its dtype is float32 and has a shape of " "`[batch_size]`. The `paths` tensor containing the highest scoring tag " "indices. Its dtype is int64 and has a shape of `[batch_size, " "sequence_length]`." msgstr ""