paper-with-me

Papers

WECHSEL: Effective initialization of subword embeddings for cross-lingual transfer of monolingual language models

2021-12-13 · NAACL 2022 7 · Benjamin Minixhofer, Fabian Paischer, Navid Rekabsaz

Large pretrained language models (LMs) have become the central building block of many NLP applications. Training these models requires ever more computational resources and most of the existing models are trained on English text only. It is exceedingly expensive to train these models in other languages. To alleviate this problem, we introduce a novel method -- called WECHSEL -- to efficiently and effectively transfer pretrained LMs to new languages. WECHSEL can be applied to any model which uses subword-based tokenization and learns an embedding for each subword. The tokenizer of the source model (in English) is replaced with a tokenizer in the target language and token embeddings are initialized such that they are semantically similar to the English tokens by utilizing multilingual static word embeddings covering English and the target language. We use WECHSEL to transfer the English RoBERTa and GPT-2 models to four languages (French, German, Chinese and Swahili). We also study the benefits of our method on very low-resource languages. WECHSEL improves over proposed methods for cross-lingual parameter transfer and outperforms models of comparable size trained from scratch with up to 64x less training effort. Our method makes training large language models for new languages more accessible and less damaging to the environment. We make our code and models publicly available.

📄 PDF Abstract BibTeX arXiv:2112.06598

Code (1)

cpjku/wechsel 공식 구현 pytorch

Tasks

Cross-Lingual TransferWord Embeddings

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Adam 설명 없음
Cosine Annealing Cosine Annealing is a type of learning rate schedule that has the effect of starting with a large learning rate that is relatively rapidly decreased to a minimum value before…
Linear Warmup With Linear Decay Linear Warmup With Linear Decay is a learning rate schedule in which we increase the learning rate linearly for $n$ updates and then linearly decay afterwards.
Residual Connection 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…

Similar Papers 제목 키워드 기반

WECHSEL: Effective initialization of subword embeddings for cross-lingual transfer of monolingual language models

2021-10-16 · ACL ARR October 2021 10 · Anonymous

Recently, large pretrained language models (LMs) have gained popularity. Training these models requires ever more computational resources and most of the existing models are trained on English text only. It is exceedingl…

Cross-Lingual TransferWord Embeddings

LGSE: Lexically Grounded Subword Embedding Initialization for Low-Resource Language Adaptation

2026-03-23 · Hailay Teklehaymanot, Dren Fazlija, Wolfgang Nejdl arxiv

Adapting pretrained language models to low-resource, morphologically rich languages remains a significant challenge. Existing vocabulary expansion methods typically rely on arbitrarily segmented subword units, resulting …

Text ClassificationQuestion Answering

Beyond Initialization Loss: A Systematic Study of Token Embedding Initialization Strategies for LLM Vocabulary Extension

2026-08-04 · Raviraj Joshi, Utkarsh Vaidya, Sanjay Singh Chauhan, Niranjan Wartikar arxiv

Vocabulary extension is an efficient way to adapt pretrained large language models (LLMs) to new languages, but the initialization of newly added token embeddings can strongly affect continued pre-training (CPT) efficien…

Semantic Retrieval

Subword-based Compact Reconstruction of Word Embeddings

2019-06-01 · NAACL 2019 6 · Shota Sasaki, Jun Suzuki, Kentaro Inui

The idea of subword-based word embeddings has been proposed in the literature, mainly for solving the out-of-vocabulary (OOV) word problem observed in standard word-based word embeddings. In this paper, we propose a meth…

Word Embeddings

Representation Learning for Unseen Words by Bridging Subwords to Semantic Networks

2020-05-01 · LREC 2020 5 · Yeachan Kim, Kang-Min Kim, SangKeun Lee

Pre-trained word embeddings are widely used in various fields. However, the coverage of pre-trained word embeddings only includes words that appeared in corpora where pre-trained embeddings are learned. It means that the…

Representation LearningWord Embeddings