# Metadata Conditioning Accelerates Language Model Pre-training

arXiv: https://arxiv.org/abs/2501.01956 · HTML: https://mlsanity.com/paper/2501.01956/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Builds on 46 of its 57 references that are in the map.
- First posted on arXiv 3 Jan 2025.

## Builds on (most cited first, 30 of 46)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering](https://mlsanity.com/paper/1809.02789/) (2018)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [Scaling Language Models: Methods, Analysis & Insights from Training Gopher](https://mlsanity.com/paper/2112.11446/) (2021)
- [PaLM 2 Technical Report](https://mlsanity.com/paper/2305.10403/) (2023)
- [The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale](https://mlsanity.com/paper/2406.17557/) (2024)
- [Multilingual Denoising Pre-training for Neural Machine Translation](https://mlsanity.com/paper/2001.08210/) (2020)
- [Bag of Tricks for Efficient Text Classification](https://mlsanity.com/paper/1607.01759/) (2016)
- [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019)
- [Deduplicating Training Data Makes Language Models Better](https://mlsanity.com/paper/2107.06499/) (2021)
- [CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data](https://mlsanity.com/paper/1911.00359/) (2019)
- [Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research](https://mlsanity.com/paper/2402.00159/) (2024)
- [Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruning](https://mlsanity.com/paper/2310.06694/) (2023)
- [Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification](https://mlsanity.com/paper/1903.04561/) (2019)
- [Scaling Data-Constrained Language Models](https://mlsanity.com/paper/2305.16264/) (2023)
- [DoReMi: Optimizing Data Mixtures Speeds Up Language Model Pretraining](https://mlsanity.com/paper/2305.10429/) (2023)
- [DataComp-LM: In search of the next generation of training sets for language models](https://mlsanity.com/paper/2406.11794/) (2024)
- [Data Selection for Language Models via Importance Resampling](https://mlsanity.com/paper/2302.03169/) (2023)
- [SemDeDup: Data-efficient learning at web-scale through semantic deduplication](https://mlsanity.com/paper/2303.09540/) (2023)
- [Time-Aware Language Models as Temporal Knowledge Bases](https://mlsanity.com/paper/2106.15110/) (2021)

## Built on by (most cited first, 22 of 22)

- [Organize the Web: Constructing Domains Enhances Pre-Training Data Curation](https://mlsanity.com/paper/2502.10341/) (2025)
- [Synthesizing Privacy-Preserving Text Data via Finetuning without Finetuning Billion-Scale LLMs](https://mlsanity.com/paper/2503.12347/) (2025)
- [PiKE: Adaptive Data Mixing for Large-Scale Multi-Task Learning Under Low Gradient Conflicts](https://mlsanity.com/paper/2502.06244/) (2025)
- [Towards Universal Offline Black-Box Optimization via Learning Language Model Embeddings](https://mlsanity.com/paper/2506.07109/) (2025)
- [AdaptMI: Adaptive Skill-based In-context Math Instruction for Small Language Models](https://mlsanity.com/paper/2505.00147/) (2025)
- [When Does Metadata Conditioning (NOT) Work for Language Model Pre-Training? A Study with Context-Free Grammars](https://mlsanity.com/paper/2504.17562/) (2025)
- [Treasure Hunt: Real-time Targeting of the Long Tail using Training-Time Markers](https://mlsanity.com/paper/2506.14702/) (2025)
- [Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models](https://mlsanity.com/paper/2506.17585/) (2025)
- [Improving Attributed Long-form Question Answering with Intent Awareness](https://mlsanity.com/paper/2603.27435/) (2026)
- [Hubs or Fringes: Pretraining Data Selection via Web Graph Centrality](https://mlsanity.com/paper/2606.11499/) (2026)
- [FIRE: Flexible Integration of Data Quality Ratings for Effective Pre-Training](https://mlsanity.com/paper/2502.00761/) (2025)
- [A Theory for Conditional Generative Modeling on Multiple Data Sources](https://mlsanity.com/paper/2502.14583/) (2025)
- [Cream of the Crop: Harvesting Rich, Scalable and Transferable Multi-Modal Data for Instruction Fine-Tuning](https://mlsanity.com/paper/2503.13383/) (2025)
- [URLs Help, Topics Guide: Understanding Metadata Utility in LLM Training](https://mlsanity.com/paper/2505.16570/) (2025)
- [Objective Matters: Fine-Tuning Objectives Shape Safety, Robustness, and Persona Drift](https://mlsanity.com/paper/2601.12639/) (2026)
- [Synthetic Mixed Training: Scaling Parametric Knowledge Acquisition Beyond RAG](https://mlsanity.com/paper/2603.23562/) (2026)
- [KoCo: Conditioning Language Model Pre-training on Knowledge Coordinates](https://mlsanity.com/paper/2604.12397/) (2026)
- [Self-Prompting Diffusion Transformer for Open-Vocabulary Scene Text Editing via In-Context Learning](https://mlsanity.com/paper/2605.15523/) (2026)
- [TokenSmith: Streamlining Data Editing, Search, and Inspection for Large-Scale Language Model Training and Interpretability](https://mlsanity.com/paper/2507.19419/) (2025)
- [MAPLE: Metadata Conditioned LLM Pretraining for Locale-Aware Question Answering](https://mlsanity.com/paper/2601.15236/) (2026)
- [SPT: Skills as Pre-Training Data for Agentic Language Models](https://mlsanity.com/paper/2608.26563/) (2026)
- [Conflicting Supervision Moves Commitment, Not Capability: A 12.29σ arrangement effect that is exactly zero under a convention-agnostic score](https://mlsanity.com/paper/2610.00234/) (2026)

## Abstract (arXiv)

The vast diversity of styles, domains, and quality levels present in language model pre-training corpora is essential in developing general model capabilities, but efficiently learning and deploying the correct behaviors exemplified in each of these heterogeneous data sources is challenging. To address this, we propose a new method, termed Metadata Conditioning then Cooldown (MeCo), to incorporate additional learning cues during pre-training. MeCo first provides metadata (e.g., URLs like www.wikipedia.org) alongside the text during training and later uses a cooldown phase with only the standard text, thereby enabling the model to function normally even without metadata. MeCo significantly accelerates pre-training across different model scales (600M to 8B parameters) and training sources (C4, RefinedWeb, and DCLM). For instance, a 1.6B language model trained with MeCo matches the downstream task performance of standard pre-training while using 33% less data. Additionally, MeCo enables us to steer language models by conditioning the inference prompt on either real or fabricated metadata that encodes the desired properties of the output: for example, prepending wikipedia.org to reduce harmful generations or factquizmaster.com (fabricated) to improve common knowledge task performance. We also demonstrate that MeCo is compatible with different types of metadata, such as model-generated topics. MeCo is remarkably simple, adds no computational overhead, and demonstrates promise in producing more capable and steerable language models.