{"version":1,"type":"story","url":"https://digestai.news/story/aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter","json":"https://digestai.news/story/aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter.json","markdown":"https://digestai.news/story/aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter.md","slug":"aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter","headline":"AQL-Lite improves Arabic LLMs with curated data and partial-parameter training","summary":"Researchers introduced **AQL-Lite**, a framework designed to efficiently adapt multilingual large language models (LLMs) to **Modern Standard Arabic (MSA)**. The method addresses challenges like noisy web data, dialectal variation, and high continual pretraining (CPT) costs by combining a quality-first data curation pipeline with partial-parameter training. This includes structural cleaning, heuristic/model-based filtering, MSA selection, and deduplication, followed by training **RS-LoRA adapters** alongside the last eight transformer blocks and embedding matrix under layer-wise learning rates.\n\nAt a **1B-token** budget, CPT on **AQL-curated data** improved performance by **3.51 points** across four Arabic benchmarks compared to raw web text, which showed no gain over the base model. In a **5B-token** comparison, AQL-Lite achieved **94% of the gain** from full-parameter CPT in **52% of the wall-clock time**, outperforming adapter-only methods on all benchmarks.","keyPoints":["AQL-Lite uses curated Arabic data and partial-parameter training to cut costs and improve MSA LLM performance","At 1B tokens, AQL-curated data boosts benchmark scores by 3.51 points over raw web text","Scaled to 40B MSA tokens, the 12B model scores 75.3 on the Open Arabic LLM Leaderboard v2"],"whyItMatters":"AQL-Lite offers a practical, resource-efficient way to train high-quality Arabic LLMs without full-parameter fine-tuning, addressing gaps in multilingual AI for Arabic speakers.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":["AQL-Lite","RS-LoRA adapters"],"people":[]},"firstPublishedAt":"2026-10-23T04:00:00Z","updatedAt":"2026-10-23T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.CL","title":"AQL-Lite: Enabling Efficient Knowledge Acquisition for Arabic LLMs via Data-Centric Continual Pretraining","url":"https://arxiv.org/abs/2610.30344","publishedAt":"2026-10-23T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"AQL-Lite improves Arabic LLMs with curated data and partial-parameter training\", 23 October 2026, https://digestai.news/story/aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter","publisher":"Digest AI","title":"AQL-Lite improves Arabic LLMs with curated data and partial-parameter training","datePublished":"2026-10-23T04:00:00Z","url":"https://digestai.news/story/aql-lite-improves-arabic-llms-with-curated-data-and-partial-parameter"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}