When Life Gives You Samples: The Benefits of Scaling up Inference Compute for Multilingual LLMs

• URL: http://arxiv.org/abs/2506.20544v1
• Date: Wed, 25 Jun 2025 15:37:53 GMT
• Title: When Life Gives You Samples: The Benefits of Scaling up Inference Compute for Multilingual LLMs
• Authors: Ammar Khairi, Daniel D'souza, Ye Shen, Julia Kreutzer, Sara Hooker,
• Abstract summary: We propose novel sampling and selection strategies specifically adapted for multilingual and multi-task inference scenarios.<n>Our combined sampling and selection methods lead to an average +6.8 jump in win-rates for our 8B models on m-Arena-v2.0 prompts.<n>Our results underscore the need for language- and task-aware approaches to inference-time compute, aiming to democratize performance improvements in underrepresented languages.
• Score: 10.406096751902899
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Recent advancements in large language models (LLMs) have shifted focus toward scaling inference-time compute, improving performance without retraining the model. A common approach is to sample multiple outputs in parallel, and select one of these as the final output. However, work to date has focused on English and a handful of domains such as math and code. In contrast, we are most interested in techniques that generalize across open-ended tasks, formally verifiable tasks, and across languages. In this work, we study how to robustly scale inference-time compute for open-ended generative tasks in a multilingual, multi-task setting. Our findings show that both sampling strategy based on temperature variation and selection strategy must be adapted to account for diverse domains and varied language settings. We evaluate existing selection methods, revealing that strategies effective in English often fail to generalize across languages. We propose novel sampling and selection strategies specifically adapted for multilingual and multi-task inference scenarios, and show they yield notable gains across languages and tasks. In particular, our combined sampling and selection methods lead to an average +6.8 jump in win-rates for our 8B models on m-ArenaHard-v2.0 prompts, against proprietary models such as Gemini. At larger scale, Command-A (111B model) equipped with our methods, shows +9.0 improvement in win-rates on the same benchmark with just five samples against single-sample decoding, a substantial increase at minimal cost. Our results underscore the need for language- and task-aware approaches to inference-time compute, aiming to democratize performance improvements in underrepresented languages.

Related papers

• SOI Matters: Analyzing Multi-Setting Training Dynamics in Pretrained Language Models via Subsets of Interest [5.882817862856554]
  This work investigates the impact of multi-task, multi-lingual, and multi-source learning approaches on the robustness and performance of pretrained language models.<n>Subsets of Interest (SOI) identifies six distinct learning behavior patterns during training, including forgettable examples, unlearned examples, and always correct examples.<n>Our results demonstrate that multi-source learning consistently improves out-of-distribution performance by up to 7%, while multi-task learning shows mixed results with notable gains in similar task combinations.
  arXiv  Detail & Related papers  (2025-07-21T04:43:21Z)
• MMTEB: Massive Multilingual Text Embedding Benchmark [85.18187649328792]
  We introduce the Massive Multilingual Text Embedding Benchmark (MMTEB)<n>MMTEB covers over 500 quality-controlled evaluation tasks across 250+ languages.<n>We develop several highly multilingual benchmarks, which we use to evaluate a representative set of models.
  arXiv  Detail & Related papers  (2025-02-19T10:13:43Z)
• P-MMEval: A Parallel Multilingual Multitask Benchmark for Consistent Evaluation of LLMs [84.24644520272835]
  We introduce P-MMEval, a large-scale benchmark covering effective fundamental and capability-specialized datasets.<n>P-MMEval delivers consistent language coverage across various datasets and provides parallel samples.<n>We conduct extensive experiments on representative multilingual model series to compare performances across models and tasks.
  arXiv  Detail & Related papers  (2024-11-14T01:29:36Z)
• Multilingual Few-Shot Learning via Language Model Retrieval [18.465566186549072]
  Transformer-based language models have achieved remarkable success in few-shot in-context learning. We conduct a study of retrieving semantically similar few-shot samples and using them as the context. We evaluate the proposed method on five natural language understanding datasets related to intent detection, question classification, sentiment analysis, and topic classification.
  arXiv  Detail & Related papers  (2023-06-19T14:27:21Z)
• Efficient Spoken Language Recognition via Multilabel Classification [53.662747523872305]
  We show that our models obtain competitive results while being orders of magnitude smaller and faster than current state-of-the-art methods. Our multilabel strategy is more robust to unseen non-target languages compared to multiclass classification.
  arXiv  Detail & Related papers  (2023-06-02T23:04:19Z)
• Beyond Contrastive Learning: A Variational Generative Model for Multilingual Retrieval [109.62363167257664]
  We propose a generative model for learning multilingual text embeddings. Our model operates on parallel data in $N$ languages. We evaluate this method on a suite of tasks including semantic similarity, bitext mining, and cross-lingual question retrieval.
  arXiv  Detail & Related papers  (2022-12-21T02:41:40Z)
• Multi Task Learning For Zero Shot Performance Prediction of Multilingual Models [12.759281077118567]
  Massively Multilingual Transformer based Language Models have been observed to be surprisingly effective on zero-shot transfer across languages. We build upon some of the existing techniques for predicting the zero-shot performance on a task, by modeling it as a multi-task learning problem.
  arXiv  Detail & Related papers  (2022-05-12T14:47:03Z)
• 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)
• A Hierarchical Model for Spoken Language Recognition [29.948719321162883]
  Spoken language recognition ( SLR) refers to the automatic process used to determine the language present in a speech sample. We propose a novel hierarchical approach were two PLDA models are trained, one to generate scores for clusters of highly related languages and a second one to generate scores conditional to each cluster. We show that this hierarchical approach consistently outperforms the non-hierarchical one for detection of highly related languages.
  arXiv  Detail & Related papers  (2022-01-04T22:10:36Z)
• CoSDA-ML: Multi-Lingual Code-Switching Data Augmentation for Zero-Shot Cross-Lingual NLP [68.2650714613869]
  We propose a data augmentation framework to generate multi-lingual code-switching data to fine-tune mBERT. Compared with the existing work, our method does not rely on bilingual sentences for training, and requires only one training process for multiple target languages.
  arXiv  Detail & Related papers  (2020-06-11T13:15:59Z)
• XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization [128.37244072182506]
  Cross-lingual TRansfer Evaluation of Multilinguals XTREME is a benchmark for evaluating the cross-lingual generalization capabilities of multilingual representations across 40 languages and 9 tasks. We demonstrate that while models tested on English reach human performance on many tasks, there is still a sizable gap in the performance of cross-lingually transferred models.
  arXiv  Detail & Related papers  (2020-03-24T19:09:37Z)

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.