Pith. sign in

REVIEW 2 cited by

Multilingual Instruction Tuning With Just a Pinch of Multilinguality

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2401.01854 v4 pith:JPHODUCS submitted 2024-01-03 cs.CL cs.AIcs.LG

Multilingual Instruction Tuning With Just a Pinch of Multilinguality

classification cs.CL cs.AIcs.LG
keywords languagesmultilingualtuningmodelsinstructioninstruction-followingduringeven
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

As instruction-tuned large language models (LLMs) gain global adoption, their ability to follow instructions in multiple languages becomes increasingly crucial. In this work, we investigate how multilinguality during instruction tuning of a multilingual LLM affects instruction-following across languages from the pre-training corpus. We first show that many languages transfer some instruction-following capabilities to other languages from even monolingual tuning. Furthermore, we find that only 40 multilingual examples integrated in an English tuning set substantially improve multilingual instruction-following, both in seen and unseen languages during tuning. In general, we observe that models tuned on multilingual mixtures exhibit comparable or superior performance in multiple languages compared to monolingually tuned models, despite training on 10x fewer examples in those languages. Finally, we find that diversifying the instruction tuning set with even just 2-4 languages significantly improves cross-lingual generalization. Our results suggest that building massively multilingual instruction-tuned models can be done with only a very small set of multilingual instruction-responses.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. IRIS: Interleaved Reinforcement with Incremental Staged Curriculum for Cross-Lingual Mathematical Reasoning

    cs.CL 2026-04 unverdicted novelty 6.0

    IRIS interleaves staged curriculum supervised fine-tuning with reverse-curriculum reinforcement learning using a composite reward to improve mathematical reasoning in English and low-resource Indian languages, accompa...

  2. English is Not All You Need: Systematically Exploring the Role of Multilinguality in LLM Post-Training

    cs.CL 2026-04 unverdicted novelty 6.0

    Systematic experiments demonstrate that multilingual coverage in LLM post-training improves results for all languages and tasks compared to English-only, with low-resource languages gaining most and zero-shot transfer...