mlsanityData to 14 Oct 2026

CLewR: Curriculum Learning with Restarts for Machine Translation Preference Learning

arXiv 2601.05858 ↗First posted 9 Jan 2026cs.CL cs.AI cs.LGcited by 1

On the map: LLM alignment / RLHF / preference optimisation → LLM preference optimization methods

Abstract (arXiv)

Large language models (LLMs) have demonstrated competitive performance in zero-shot multilingual machine translation (MT). Some follow-up works further improved MT performance via preference optimization, but they leave a key aspect largely underexplored: the order in which data samples are given during training. We address this topic by integrating curriculum learning into various state-of-the-art preference optimization algorithms to boost MT performance. We introduce a novel curriculum learning strategy with restarts (CLewR), which reiterates easy-to-hard curriculum multiple times during training to effectively mitigate the catastrophic forgetting of easy examples. We demonstrate consistent gains across several model families (Gemma2, Qwen2.5, Llama3.1) and preference optimization techniques. We publicly release our code at https://github.com/alexandra-dragomir/CLewR.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 22 of its 43 references that are in the map. First posted on arXiv 9 Jan 2026.

Builds on

22 of 43 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences12 Mar
  2. 2024Smaug: Fixing Failure Modes of Preference Optimisation with DPO-Positive20 Feb
  3. 2024Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation16 Jan

Other references · 19

  1. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  4. 2022No Language Left Behind: Scaling Human-Centered Machine Translationcited by 1,312
  5. 2024Tower: An Open Multilingual Large Language Model for Translation-Related Taskscited by 206
  6. 2019Competence-based Curriculum Learning for Neural Machine Translationcited by 202
  7. 2017Reverse Curriculum Generation for Reinforcement Learningcited by 198
  8. 2017Curriculum Learning and Minibatch Bucketing in Neural Machine Translationcited by 63
  9. 2020Norm-Based Curriculum Learning for Neural Machine Translationcited by 60
  10. 2025Multilingual Machine Translation with Open Large Language Models at Practical Scale: An Empirical Studycited by 38
  11. 2024Curriculum Direct Preference Optimization for Diffusion and Consistency Modelscited by 36
  12. 2024X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scalecited by 36
7 more
  1. 2021Does the Order of Training Samples Matter? Improving Neural Data-to-Text Generation with Curriculum Learningcited by 34
  2. 2021Meta-Curriculum Learning for Domain Adaptation in Neural Machine Translationcited by 28
  3. 2023Dynamic Masking Rate Schedules for MLM Pretrainingcited by 12
  4. 2024A Novel Cartography-Based Curriculum Learning Method Applied on RoNLI: The First Romanian Natural Language Inference Corpuscited by 8
  5. 2021Density-Based Dynamic Curriculum Learning for Intent Detectioncited by 7
  6. 2020Task-Level Curriculum Learning for Non-Autoregressive Neural Machine Translationcited by 3
  7. 2025Task-Informed Anti-Curriculum by Masking Improves Downstream Performance on Textcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026XekRung Technical Reportcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Bridging the Linguistic Divide: A Survey on Leveraging Large Language Models for Machine Translation
  2. 2024Making LLMs Better Many-to-Many Speech-to-Text Translators with Curriculum Learning
  3. 2025RL from Teacher-Model Refinement: Gradual Imitation Learning for Machine Translation
  4. 2026Qwen-MT-Lumen at WMT2026: Instruction-Specialized Translation via Quality-Aware Training and On-Policy Distillation
  5. 2025The Art of Asking: Multilingual Prompt Optimization for Synthetic Data
  6. 2025M2PO: Multi-Perspective Multi-Pair Preference Optimization for Machine Translation
  7. 2025Preference Curriculum: LLMs Should Always Be Pretrained on Their Preferred Data
  8. 2026Cross-Preference Learning for Sentence-Level and Context-Aware Machine Translation