Research map / Large language models / Language, safety and interpretability
Toward Robust Multilingual Adaptation of LLMs for Low-Resource Languages
Haolin Li, Haipeng Zhang, Mang Li, Yaohua Wang, Lijie Wen, zhang yu, Biqing Huang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add the LLM critic (ablation) | LiRA · LazRetrieval | nDCG@10: 71.29 -> 77.71 | Improved |
| add the Embeds critic (ablation) | LiRA · LazRetrieval | nDCG@10: 65.77 -> 77.71 | Improved |
| add the Embeds critic (ablation) | LiRA · STS22 | Pearson: 61.78 -> 75.00 | Improved |
| add translations (ablation) | LiRA · LazRetrieval | nDCG@10: 75.48 -> 77.71 | Improved |
| add multilingual encoder (ablation) | LiRA · STS22 | Pearson: 72.43 -> 75.00 | Improved |
| add the FIFO queue (ablation) | LiRA · LazRetrieval | nDCG@10: 64.29 -> 77.71 | Improved |
| add the LLM critic (ablation) | LiRA · MGSM | accuracy (%): 68.9 -> 71.1 | Improved |
| add LiRA-Large to Qwen3-Embedding-8B | Qwen3-E-8B · LazRetrieval | nDCG@10 average: 68.05 -> 72.86 | Improved |
| scale LiRA from Base to Large to Max | Qwen3-E-8B · LazRetrieval | nDCG@10 average: 69.61 -> 72.86 -> 77.71 | Improved |
| add LiRA to Qwen3-Embedding-8B | Qwen3-E-8B · Public retrieval | average score: 79.57 -> 81.35 | Improved |
| add LiRA-Large to Qwen3-8B | Qwen3-8B · MGSM | accuracy (%): 69.4 -> 71.1 | Improved |
| add LiRA-Large to Qwen3-8B | Qwen3-8B · X-CSQA | accuracy (%): 62.9 -> 65.5 | Improved |
Closest papers by meaning
- Language Ranker: A Metric for Quantifying LLM Performance Across High and Low-Resource Languages (AAAI 2025 · similarity 0.89)
- QA-LoRA: Quantization-Aware Low-Rank Adaptation of Large Language Models (ICLR 2024 · similarity 0.88)
- Exploring Polyglot Harmony: On Multilingual Data Allocation for Large Language Models Pretraining (NeurIPS 2025 · similarity 0.88)
- Enhancing Multilingual LLM Pretraining with Model-Based Data Selection (NeurIPS 2025 · similarity 0.88)
- LLM Alignment as Retriever Optimization: An Information Retrieval Perspective (ICML 2025 · similarity 0.88)
- Multilingual LLMs Inherently Reward In-Language Time-Sensitive Semantic Alignment for Low-Resource Languages (AAAI 2025 · similarity 0.87)
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.
