Beyond Random Sampling: Efficient Language Model Pretraining via Curriculum Learning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Yang, Mohamed, Amr, Abdine, Hadi, Shang, Guokan, Vazirgiannis, Michalis
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918311408697344
author Zhang, Yang
Mohamed, Amr
Abdine, Hadi
Shang, Guokan
Vazirgiannis, Michalis
author_facet Zhang, Yang
Mohamed, Amr
Abdine, Hadi
Shang, Guokan
Vazirgiannis, Michalis
contents Curriculum learning-organizing training data from easy to hard-has improved efficiency across machine learning domains, yet remains underexplored for language model pretraining. We present the first systematic investigation of curriculum learning in LLM pretraining, with over 200 models trained on up to 100B tokens across three strategies: vanilla curriculum learning, pacing-based sampling, and interleaved curricula, guided by six difficulty metrics spanning linguistic and information-theoretic properties. We evaluate performance on eight benchmarks under three realistic scenarios: limited data, unlimited data, and continual training. Our experiments show that curriculum learning consistently accelerates convergence in early and mid-training phases,reducing training steps by $18-45\%$ to reach baseline performance. When applied as a warmup strategy before standard random sampling, curriculum learning yields sustained improvements up to $3.5\%$. We identify compression ratio, lexical diversity (MTLD), and readability (Flesch Reading Ease) as the most effective difficulty signals. Our findings demonstrate that data ordering-orthogonal to existing data selection methods-provides a practical mechanism for more efficient LLM pretraining.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11300
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Random Sampling: Efficient Language Model Pretraining via Curriculum Learning
Zhang, Yang
Mohamed, Amr
Abdine, Hadi
Shang, Guokan
Vazirgiannis, Michalis
Computation and Language
Artificial Intelligence
Curriculum learning-organizing training data from easy to hard-has improved efficiency across machine learning domains, yet remains underexplored for language model pretraining. We present the first systematic investigation of curriculum learning in LLM pretraining, with over 200 models trained on up to 100B tokens across three strategies: vanilla curriculum learning, pacing-based sampling, and interleaved curricula, guided by six difficulty metrics spanning linguistic and information-theoretic properties. We evaluate performance on eight benchmarks under three realistic scenarios: limited data, unlimited data, and continual training. Our experiments show that curriculum learning consistently accelerates convergence in early and mid-training phases,reducing training steps by $18-45\%$ to reach baseline performance. When applied as a warmup strategy before standard random sampling, curriculum learning yields sustained improvements up to $3.5\%$. We identify compression ratio, lexical diversity (MTLD), and readability (Flesch Reading Ease) as the most effective difficulty signals. Our findings demonstrate that data ordering-orthogonal to existing data selection methods-provides a practical mechanism for more efficient LLM pretraining.
title Beyond Random Sampling: Efficient Language Model Pretraining via Curriculum Learning
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2506.11300