# Does your data spark joy? Performance gains from domain upsampling at the end of training

arXiv: https://arxiv.org/abs/2406.03476 · HTML: https://mlsanity.com/paper/2406.03476/

## Key facts

- Cited by 52 papers in the map (data through 14 Oct 2026).
- Builds on 26 of its 36 references that are in the map.
- First posted on arXiv 5 Jun 2024.

## Builds on (most cited first, 26 of 26)

- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [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)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [Emergent Abilities of Large Language Models](https://mlsanity.com/paper/2206.07682/) (2022)
- [BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions](https://mlsanity.com/paper/1905.10044/) (2019)
- [Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering](https://mlsanity.com/paper/1809.02789/) (2018)
- [Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models](https://mlsanity.com/paper/2206.04615/) (2022)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation](https://mlsanity.com/paper/2108.12409/) (2021)
- [Scaling Vision Transformers](https://mlsanity.com/paper/2106.04560/) (2021)
- [The LAMBADA dataset: Word prediction requiring a broad discourse context](https://mlsanity.com/paper/1606.06031/) (2016)
- [Are NLP Models really able to Solve Simple Math Word Problems?](https://mlsanity.com/paper/2103.07191/) (2021)
- [The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Only](https://mlsanity.com/paper/2306.01116/) (2023)
- [AGIEval: A Human-Centric Benchmark for Evaluating Foundation Models](https://mlsanity.com/paper/2304.06364/) (2023)
- [CoQA: A Conversational Question Answering Challenge](https://mlsanity.com/paper/1808.07042/) (2018)
- [Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research](https://mlsanity.com/paper/2402.00159/) (2024)
- [Symbolic Discovery of Optimization Algorithms](https://mlsanity.com/paper/2302.06675/) (2023)
- [DoReMi: Optimizing Data Mixtures Speeds Up Language Model Pretraining](https://mlsanity.com/paper/2305.10429/) (2023)

## Built on by (most cited first, 30 of 52)

- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models](https://mlsanity.com/paper/2502.18443/) (2025)
- [Reuse, Don't Retrain: A Recipe for Continued Pretraining of Language Models](https://mlsanity.com/paper/2407.07263/) (2024)
- [Nemotron-CLIMB: CLustering-based Iterative Data Mixture Bootstrapping for Language Model Pre-training](https://mlsanity.com/paper/2504.13161/) (2025)
- [Diffusion Language Models are Super Data Learners](https://mlsanity.com/paper/2511.03276/) (2025)
- [Nemotron-CC-Math: A 133 Billion-Token-Scale High Quality Math Pretraining Dataset](https://mlsanity.com/paper/2508.15096/) (2025)
- [Aioli: A Unified Optimization Framework for Language Model Data Mixing](https://mlsanity.com/paper/2411.05735/) (2024)
- [Steering off Course: Reliability Challenges in Steering Language Models](https://mlsanity.com/paper/2504.04635/) (2025)
- [EuroBERT: Scaling Multilingual Encoders for European Languages](https://mlsanity.com/paper/2503.05500/) (2025)
- [Llama-3.1-FoundationAI-SecurityLLM-Base-8B Technical Report](https://mlsanity.com/paper/2504.21039/) (2025)
- [Maximize Your Data's Potential: Enhancing LLM Accuracy with Two-Phase Pretraining](https://mlsanity.com/paper/2412.15285/) (2024)
- [Signal and Noise: A Framework for Reducing Uncertainty in Language Model Evaluation](https://mlsanity.com/paper/2508.13144/) (2025)
- [Encode, Think, Decode: Scaling test-time reasoning with recursive latent thoughts](https://mlsanity.com/paper/2510.07358/) (2025)
- [The Zamba2 Suite: Technical Report](https://mlsanity.com/paper/2411.15242/) (2024)
- [A Survey on LLM Mid-Training](https://mlsanity.com/paper/2510.23081/) (2025)
- [Language Models Improve When Pretraining Data Matches Target Tasks](https://mlsanity.com/paper/2507.12466/) (2025)
- [Data Mixing Can Induce Phase Transitions in Knowledge Acquisition](https://mlsanity.com/paper/2505.18091/) (2025)
- [FlexOlmo: Open Language Models for Flexible Data Use](https://mlsanity.com/paper/2507.07024/) (2025)
- [Zyda-2: a 5 Trillion Token High-Quality Dataset](https://mlsanity.com/paper/2411.06068/) (2024)
- [Replaying pre-training data improves fine-tuning](https://mlsanity.com/paper/2603.04964/) (2026)
- [Midtraining Bridges Pretraining and Posttraining Distributions](https://mlsanity.com/paper/2510.14865/) (2025)
- [MixMin: Finding Data Mixtures via Convex Minimization](https://mlsanity.com/paper/2502.10510/) (2025)
- [Arcee Trinity Large Technical Report](https://mlsanity.com/paper/2602.17004/) (2026)
- [Predicting Emergent Capabilities by Finetuning](https://mlsanity.com/paper/2411.16035/) (2024)
- [Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework](https://mlsanity.com/paper/2503.21023/) (2025)
- [The Finetuner's Fallacy: When to Pretrain with Your Finetuning Data](https://mlsanity.com/paper/2603.16177/) (2026)
- [HyperCLOVA X THINK Technical Report](https://mlsanity.com/paper/2506.22403/) (2025)
- [When Should We Introduce Safety Interventions During Pretraining?](https://mlsanity.com/paper/2601.07087/) (2026)
- [From Growing to Looping: A Unified View of Iterative Computation in LLMs](https://mlsanity.com/paper/2602.16490/) (2026)
- [Covenant-72B: Pre-Training a 72B LLM with Trustless Peers Over-the-Internet](https://mlsanity.com/paper/2603.08163/) (2026)

## Abstract (arXiv)

Pretraining datasets for large language models (LLMs) have grown to trillions of tokens composed of large amounts of CommonCrawl (CC) web scrape along with smaller, domain-specific datasets. It is expensive to understand the impact of these domain-specific datasets on model capabilities as training at large FLOP scales is required to reveal significant changes to difficult and emergent benchmarks. Given the increasing cost of experimenting with pretraining data, how does one determine the optimal balance between the diversity in general web scrapes and the information density of domain specific data? In this work, we show how to leverage the smaller domain specific datasets by upsampling them relative to CC at the end of training to drive performance improvements on difficult benchmarks. This simple technique allows us to improve up to 6.90 pp on MMLU, 8.26 pp on GSM8K, and 6.17 pp on HumanEval relative to the base data mix for a 7B model trained for 1 trillion (T) tokens, thus rivaling Llama-2 (7B)unicodex2014a model trained for twice as long. We experiment with ablating the duration of domain upsampling from 5% to 30% of training and find that 10% to 20% percent is optimal for navigating the tradeoff between general language modeling capabilities and targeted benchmarks. We also use domain upsampling to characterize at scale the utility of individual datasets for improving various benchmarks by removing them during this final phase of training. This tool opens up the ability to experiment with the impact of different pretraining datasets at scale, but at an order of magnitude lower cost compared to full pretraining runs.