paper-with-me

홈 › Papers

The emergence of number and syntax units in LSTM language models

2019-03-18 · NAACL 2019 6 · Yair Lakretz, German Kruszewski, Theo Desbordes, Dieuwke Hupkes, Stanislas Dehaene, Marco Baroni

Recent work has shown that LSTMs trained on a generic language modeling objective capture syntax-sensitive generalizations such as long-distance number agreement. We have however no mechanistic understanding of how they accomplish this remarkable feat. Some have conjectured it depends on heuristics that do not truly take hierarchical structure into account. We present here a detailed study of the inner mechanics of number tracking in LSTMs at the single neuron level. We discover that long-distance number information is largely managed by two `number units'. Importantly, the behaviour of these units is partially controlled by other units independently shown to track syntactic structure. We conclude that LSTMs are, to some extent, implementing genuinely syntactic processing mechanisms, paving the way to a more general understanding of grammatical encoding in LSTMs.

📄 PDF Abstract BibTeX arXiv:1903.07435

Code (1)

FAIRNS/Number_and_syntax_units_in_LSTM_LMs 공식 구현 pytorch

Tasks

Language ModelingLanguage Modelling

Similar Papers 제목 키워드 기반

Compositional Neural Machine Translation by Removing the Lexicon from Syntax

2020-02-06 · Tristan Thrush

The meaning of a natural language utterance is largely determined from its syntax and words. Additionally, there is evidence that humans process an utterance by separating knowledge about the lexicon from syntax knowledg…

DecoderMachine TranslationSemantic ParsingTranslation

LSTMs Can Learn Syntax-Sensitive Dependencies Well, But Modeling Structure Makes Them Better

2018-07-01 · ACL 2018 7 · Adhiguna Kuncoro, Chris Dyer, John Hale, Dani Yogatama 외

Language exhibits hierarchical structure, but recent work using a subject-verb agreement diagnostic argued that state-of-the-art language models, LSTMs, fail to learn long-range syntax sensitive dependencies. Using the s…

DiagnosticLanguage ModelingLanguage ModellingMachine Translation+1

Assessing the Ability of LSTMs to Learn Syntax-Sensitive Dependencies

2016-11-04 · TACL 2016 1 · Tal Linzen, Emmanuel Dupoux, Yoav Goldberg

The success of long short-term memory (LSTM) neural networks in language processing is typically attributed to their ability to capture long-distance statistical regularities. Linguistic regularities are often sensitive …

Language ModelingLanguage Modelling

Chromatic transitions in the emergence of syntax networks

2018-07-24

The emergence of syntax during childhood is a remarkable example of how complex correlations unfold in nonlinear ways through development. In particular, rapid transitions seem to occur as children reach the age of two, …

Language Acquisition

A Sequential Neural Encoder with Latent Structured Description for Modeling Sentences

2017-11-15 · Yu-Ping Ruan, Qian Chen, Zhen-Hua Ling

In this paper, we propose a sequential neural encoder with latent structured description (SNELSD) for modeling sentences. This model introduces latent chunk-level representations into conventional sequential neural encod…

ChunkingNatural Language InferenceSentenceSentence Embeddings+1