AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

LinguaMap: Which Layers of LLMs Speak Your Language and How to Tune Them?

J. Ben Tamo, Daniel Carlander-Reuterfelt, Jonathan Rubin, Oleg Poliannikov, Dezhi Hong, Mingxian Wang

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Efficient LLM fine-tuning systems · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fine-tune only the final layers (selective SFT)Qwen-3-32B · MMLU non-Business (code-switched)language consistency: 8.35 -> 99.62Improved
fine-tune only the final layers (selective SFT)Qwen-3-32B · MMLU non-Business (code-switched)task performance: 60.51 -> 74.44Improved
fine-tune only the final layers (selective SFT)Qwen-3-32B · XQuAD (English distractor)language consistency: 41.99 -> 97.62Improved
fine-tune only the final layers (selective SFT)Qwen-3-32B · XQuAD (English distractor)task performance: 15.81 -> 18.05No clear effect
selective SFT instead of full-scope SFTQwen-3-32B · XQuAD (English distractor)language consistency: 75.99 -> 97.62Improved
selective SFT instead of full-scope SFTQwen-3-32B · MGSM (monolingual)task performance: 90.53 -> 86.80Hurt
select final layers rather than a random layer subset (ablation)Qwen-3-32B · MGSM (monolingual)task performance: 0.13 -> 86.80Improved
select final layers rather than a random layer subset (ablation)Qwen-3-32B · MGSM (monolingual)language consistency: 65.87 -> 99.20Improved
select final layers rather than a random layer subset (ablation)Bloom-7.1B · MGSM (monolingual)language consistency: 69.47 -> 100.00Improved
fine-tune only the final layers (selective SFT)Bloom-7.1B · MMLU non-Business (code-switched)language consistency: 29.49 -> 98.66Improved
switch from monolingual to code-switched promptsQwen-3-32B · MMLUlanguage consistency: 45.17 -> 8.35Hurt
switch from monolingual to code-switched promptsQwen-3-32B · MMLUtask accuracy: 51.77 -> 60.5Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy