SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignment

• URL: http://arxiv.org/abs/2501.03681v1
• Date: Tue, 07 Jan 2025 10:29:43 GMT
• Title: SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignment
• Authors: Yuchun Fan, Yongyu Mu, Yilin Wang, Lei Huang, Junhao Ruan, Bei Li, Tong Xiao, Shujian Huang, Xiaocheng Feng, Jingbo Zhu,
• Abstract summary: We propose an efficient multilingual reasoning alignment approach that precisely identifies and fine-tunes the layers responsible for handling multilingualism. Experimental results show that our method, SLAM, only tunes 6 layers' feed-forward sub-layers including 6.5-8% of all parameters within 7B and 13B LLMs.
• Score: 78.4550589538805
• License:
• Abstract: Despite the significant improvements achieved by large language models (LLMs) in English reasoning tasks, these models continue to struggle with multilingual reasoning. Recent studies leverage a full-parameter and two-stage training paradigm to teach models to first understand non-English questions and then reason. However, this method suffers from both substantial computational resource computing and catastrophic forgetting. The fundamental cause is that, with the primary goal of enhancing multilingual comprehension, an excessive number of irrelevant layers and parameters are tuned during the first stage. Given our findings that the representation learning of languages is merely conducted in lower-level layers, we propose an efficient multilingual reasoning alignment approach that precisely identifies and fine-tunes the layers responsible for handling multilingualism. Experimental results show that our method, SLAM, only tunes 6 layers' feed-forward sub-layers including 6.5-8% of all parameters within 7B and 13B LLMs, achieving superior average performance than all strong baselines across 10 languages. Meanwhile, SLAM only involves one training stage, reducing training time by 4.1-11.9 compared to the two-stage method.

Related papers

• Demystifying Multilingual Chain-of-Thought in Process Reward Modeling [71.12193680015622]
  We tackle the challenge of extending process reward models (PRMs) to multilingual settings. We train multilingual PRMs on a dataset spanning seven languages, which is translated from English. Our results highlight the sensitivity of multilingual PRMs to both the number of training languages and the volume of English data.
  arXiv  Detail & Related papers  (2025-02-18T09:11:44Z)
• A Comprehensive Evaluation of Large Language Models on Mental Illnesses in Arabic Context [0.9074663948713616]
  Mental health disorders pose a growing public health concern in the Arab world. This study comprehensively evaluates 8 large language models (LLMs) on diverse mental health datasets.
  arXiv  Detail & Related papers  (2025-01-12T16:17:25Z)
• Inductive Linguistic Reasoning with Large Language Models [0.0]
  We investigate the abilities of large language models to perform abstract multilingual reasoning through the lens of linguistic puzzles. We employ a two-stage procedure, first generating analogical exemplars with a language model, and then applying them in-context. Our results on the modeLing dataset show that analogical prompting is effective in eliciting models' knowledge of language grammar similarities.
  arXiv  Detail & Related papers  (2024-12-09T03:37:11Z)
• The Power of Question Translation Training in Multilingual Reasoning: Broadened Scope and Deepened Insights [108.40766216456413]
  We propose a question alignment framework to bridge the gap between large language models' English and non-English performance. Experiment results show it can boost multilingual performance across diverse reasoning scenarios, model families, and sizes. We analyze representation space, generated response and data scales, and reveal how question translation training strengthens language alignment within LLMs.
  arXiv  Detail & Related papers  (2024-05-02T14:49:50Z)
• No Train but Gain: Language Arithmetic for training-free Language Adapters enhancement [59.37775534633868]
  We introduce a novel method called language arithmetic, which enables training-free post-processing. The effectiveness of the proposed solution is demonstrated on three downstream tasks in a MAD-X-based set of cross-lingual schemes.
  arXiv  Detail & Related papers  (2024-04-24T08:52:40Z)
• Breaking Language Barriers in Multilingual Mathematical Reasoning: Insights and Observations [59.056367787688146]
  This paper pioneers exploring and training powerful Multilingual Math Reasoning (xMR) LLMs. We construct the first multilingual math reasoning instruction dataset, MGSM8KInstruct, encompassing ten distinct languages. By utilizing translation, we construct the first multilingual math reasoning instruction dataset, MGSM8KInstruct, encompassing ten distinct languages.
  arXiv  Detail & Related papers  (2023-10-31T08:09:20Z)
• Tokenizer Choice For LLM Training: Negligible or Crucial? [30.33170936148845]
  We study the influence of tokenizer choice on Large Language Models (LLMs) downstream performance by training 24 mono- and multilingual LLMs. We find that the tokenizer choice can significantly impact the model's downstream performance and training costs. We show that multilingual tokenizers trained on the five most frequent European languages require vocabulary size increases of factor three in comparison to English.
  arXiv  Detail & Related papers  (2023-10-12T22:44:19Z)
• Toward Efficient Language Model Pretraining and Downstream Adaptation via Self-Evolution: A Case Study on SuperGLUE [203.65227947509933]
  This report describes our JDExplore d-team's Vega v2 submission on the SuperGLUE leaderboard. SuperGLUE is more challenging than the widely used general language understanding evaluation (GLUE) benchmark, containing eight difficult language understanding tasks.
  arXiv  Detail & Related papers  (2022-12-04T15:36:18Z)
• Por Qu\'e N\~ao Utiliser Alla Spr{\aa}k? Mixed Training with Gradient Optimization in Few-Shot Cross-Lingual Transfer [2.7213511121305465]
  We propose a one-step mixed training method that trains on both source and target data. We use one model to handle all target languages simultaneously to avoid excessively language-specific models. Our proposed method achieves state-of-the-art performance on all tasks and outperforms target-adapting by a large margin.
  arXiv  Detail & Related papers  (2022-04-29T04:05:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.