Activation steering directions for figurative categories transfer across languages in multilingual LLMs, providing evidence for reusable cross-lingual signals.
Serge Sharoff, Reinhard Rapp, and Pierre Zweigenbaum
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
abstract
We present an approach based on multilingual sentence embeddings to automatically extract parallel sentences from the content of Wikipedia articles in 85 languages, including several dialects or low-resource languages. We do not limit the the extraction process to alignments with English, but systematically consider all possible language pairs. In total, we are able to extract 135M parallel sentences for 1620 different language pairs, out of which only 34M are aligned with English. This corpus of parallel sentences is freely available at https://github.com/facebookresearch/LASER/tree/master/tasks/WikiMatrix. To get an indication on the quality of the extracted bitexts, we train neural MT baseline systems on the mined data only for 1886 languages pairs, and evaluate them on the TED corpus, achieving strong BLEU scores for many language pairs. The WikiMatrix bitexts seem to be particularly interesting to train MT systems between distant languages without the need to pivot through English.
fields
cs.CL 3years
2026 3verdicts
UNVERDICTED 3representative citing papers
SLAP is a new batch-aware pruning framework that uses distribution-aware stratified sampling and Hessian-approximated gradients to select data, claiming 20-40% less data while matching or exceeding full-dataset performance on LLM instruction tuning tasks.
The authors built and publicly released sentence-aligned simplification corpora for five languages by processing crowd-sourced data from comparable documents.
citing papers explorer
-
Cross-Lingual Steering for Figurative Language Generation
Activation steering directions for figurative categories transfer across languages in multilingual LLMs, providing evidence for reusable cross-lingual signals.
-
SLAP: Stratified Loss-based Pruning for On-Policy Data-Efficient Instruction Tuning
SLAP is a new batch-aware pruning framework that uses distribution-aware stratified sampling and Hessian-approximated gradients to select data, claiming 20-40% less data while matching or exceeding full-dataset performance on LLM instruction tuning tasks.
-
Align and Shine: Building High-Quality Sentence-Aligned Corpora for Multilingual Text Simplification
The authors built and publicly released sentence-aligned simplification corpora for five languages by processing crowd-sourced data from comparable documents.