Optimizing Pre-Training Data Mixtures with Mixtures of Data Expert Models

• URL: http://arxiv.org/abs/2502.15950v1
• Date: Fri, 21 Feb 2025 21:27:48 GMT
• Title: Optimizing Pre-Training Data Mixtures with Mixtures of Data Expert Models
• Authors: Lior Belenki, Alekh Agarwal, Tianze Shi, Kristina Toutanova,
• Abstract summary: We propose a method to optimize language model pre-training data mixtures through efficient approximation of the cross-entropy loss corresponding to each candidate mixture.<n>We use this approximation as a source of additional features in a regression model, trained from observations of model loss for a small number of mixtures.
• Score: 24.396525123797073
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: We propose a method to optimize language model pre-training data mixtures through efficient approximation of the cross-entropy loss corresponding to each candidate mixture via a Mixture of Data Experts (MDE). We use this approximation as a source of additional features in a regression model, trained from observations of model loss for a small number of mixtures. Experiments with Transformer decoder-only language models in the range of 70M to 1B parameters on the SlimPajama dataset show that our method achieves significantly better performance than approaches that train regression models using only the mixture rates as input features. Combining this improved optimization method with an objective that takes into account cross-entropy on end task data leads to superior performance on few-shot downstream evaluations. We also provide theoretical insights on why aggregation of data expert predictions can provide good approximations to model losses for data mixtures.

Related papers

• DUPRE: Data Utility Prediction for Efficient Data Valuation [49.60564885180563]
  Cooperative game theory-based data valuation, such as Data Shapley, requires evaluating the data utility and retraining the ML model for multiple data subsets. Our framework, textttDUPRE, takes an alternative yet complementary approach that reduces the cost per subset evaluation by predicting data utilities instead of evaluating them by model retraining. Specifically, given the evaluated data utilities of some data subsets, textttDUPRE fits a emphGaussian process (GP) regression model to predict the utility of every other data subset.
  arXiv  Detail & Related papers  (2025-02-22T08:53:39Z)
• DUET: Optimizing Training Data Mixtures via Feedback from Unseen Evaluation Tasks [40.91931801667421]
  This paper presents a novel global-to-local algorithm called DUET that can exploit the feedback loop by interleaving a data selection method with Bayesian optimization.<n>As a result, DUET can efficiently refine the training data mixture from a pool of data domains to maximize the model's performance on the unseen evaluation task.
  arXiv  Detail & Related papers  (2025-02-01T01:52:32Z)
• Optimizing Pretraining Data Mixtures with LLM-Estimated Utility [52.08428597962423]
  Large Language Models improve with increasing amounts of high-quality training data.<n>We find token-counts outperform manual and learned mixes, indicating that simple approaches for dataset size and diversity are surprisingly effective.<n>We propose two complementary approaches: UtiliMax, which extends token-based $200s by incorporating utility estimates from reduced-scale ablations, achieving up to a 10.6x speedup over manual baselines; and Model Estimated Data Utility (MEDU), which leverages LLMs to estimate data utility from small samples, matching ablation-based performance while reducing computational requirements by $simx.
  arXiv  Detail & Related papers  (2025-01-20T21:10:22Z)
• Scalable Data Ablation Approximations for Language Models through Modular Training and Merging [27.445079398772904]
  We propose an efficient method for approximating data ablations which trains individual models on subsets of a training corpus. We find that, given an arbitrary evaluation set, the perplexity score of a single model trained on a candidate set of data is strongly correlated with perplexity scores of parameter averages of models trained on distinct partitions of that data.
  arXiv  Detail & Related papers  (2024-10-21T06:03:49Z)
• BiMix: A Bivariate Data Mixing Law for Language Model Pretraining [47.77701041534746]
  The impact of pretraining data composition on model performance remains poorly understood. $textbfBiMix$ provides a systematic framework for understanding and optimizing data mixtures. Our work contributes both theoretical insights into data mixing dynamics and practical tools for enhancing LLM training efficiency.
  arXiv  Detail & Related papers  (2024-05-23T09:44:02Z)
• Data Mixing Laws: Optimizing Data Mixtures by Predicting Language Modeling Performance [55.872926690722714]
  We study the predictability of model performance regarding the mixture proportions in function forms. We propose nested use of the scaling laws of training steps, model sizes, and our data mixing law. Our method effectively optimize the training mixture of a 1B model trained for 100B tokens in RedPajama.
  arXiv  Detail & Related papers  (2024-03-25T17:14:00Z)
• Efficient Online Data Mixing For Language Model Pre-Training [101.45242332613944]
  Existing data selection methods suffer from slow and computationally expensive processes. Data mixing, on the other hand, reduces the complexity of data selection by grouping data points together. We develop an efficient algorithm for Online Data Mixing (ODM) that combines elements from both data selection and data mixing.
  arXiv  Detail & Related papers  (2023-12-05T00:42:35Z)
• Learning from aggregated data with a maximum entropy model [73.63512438583375]
  We show how a new model, similar to a logistic regression, may be learned from aggregated data only by approximating the unobserved feature distribution with a maximum entropy hypothesis. We present empirical evidence on several public datasets that the model learned this way can achieve performances comparable to those of a logistic model trained with the full unaggregated data.
  arXiv  Detail & Related papers  (2022-10-05T09:17:27Z)
• An Extended Multi-Model Regression Approach for Compressive Strength Prediction and Optimization of a Concrete Mixture [0.0]
  A model based evaluation of concrete compressive strength is of high value, both for the purpose of strength prediction and the mixture optimization. We take a further step towards improving the accuracy of the prediction model via the weighted combination of multiple regression methods. A proposed (GA)-based mixture optimization is proposed, building on the obtained multi-regression model.
  arXiv  Detail & Related papers  (2021-06-13T16:10:32Z)
• A similarity-based Bayesian mixture-of-experts model [0.5156484100374058]
  We present a new non-parametric mixture-of-experts model for multivariate regression problems. Using a conditionally specified model, predictions for out-of-sample inputs are based on similarities to each observed data point. Posterior inference is performed on the parameters of the mixture as well as the distance metric.
  arXiv  Detail & Related papers  (2020-12-03T18:08:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.