Analyzing Similarity Metrics for Data Selection for Language Model Pretraining

• URL: http://arxiv.org/abs/2502.02494v3
• Date: Tue, 21 Oct 2025 05:13:12 GMT
• Title: Analyzing Similarity Metrics for Data Selection for Language Model Pretraining
• Authors: Dylan Sam, Ayan Chakrabarti, Afshin Rostamizadeh, Srikumar Ramalingam, Gui Citovsky, Sanjiv Kumar,
• Abstract summary: Measuring similarity between training examples is critical for curating high-quality and diverse pretraining datasets for language models.<n>Standard off-the-shelf embedding models are not well-suited for the pretraining data curation setting.
• Score: 39.02299450717135
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Measuring similarity between training examples is critical for curating high-quality and diverse pretraining datasets for language models. However, similarity is typically computed with a generic off-the-shelf embedding model that has been trained for tasks such as retrieval. Whether these embedding-based similarity metrics are well-suited for pretraining data selection remains largely unexplored. In this paper, we propose a new framework to assess the suitability of a similarity metric specifically for data curation in language model pretraining applications. Our framework's first evaluation criterion captures how well distances reflect generalization in pretraining loss between different training examples. Next, we use each embedding model to guide a standard diversity-based data curation algorithm and measure its utility by pretraining a language model on the selected data and evaluating downstream task performance. Finally, we evaluate the capabilities of embeddings to distinguish between examples from different data sources. With these evaluations, we demonstrate that standard off-the-shelf embedding models are not well-suited for the pretraining data curation setting, underperforming even remarkably simple embeddings that are extracted from models trained on the same pretraining corpus. Our experiments are performed on the Pile, for pretraining a 1.7B parameter language model on 200B tokens. We believe our analysis and evaluation framework serves as a foundation for the future design of embeddings that specifically reason about similarity in pretraining datasets.

Related papers

• Bayesian Test-Time Adaptation for Vision-Language Models [51.93247610195295]
  Test-time adaptation with pre-trained vision-language models, such as CLIP, aims to adapt the model to new, potentially out-of-distribution test data.<n>We propose a novel approach, textbfBayesian textbfClass textbfAdaptation (BCA), which in addition to continuously updating class embeddings to adapt likelihood, also uses the posterior of incoming samples to continuously update the prior for each class embedding.
  arXiv  Detail & Related papers  (2025-03-12T10:42:11Z)
• The interplay between domain specialization and model size: a case study in the legal domain [8.653321928148547]
  We investigate the interplay between domain and model size during continual pre-training under compute-constrained scenarios.<n>Our goal is to identify a compute-efficient training regime for this scenario.<n>As model size increases, the compute-effectiveness gap between specialized and general models widens.
  arXiv  Detail & Related papers  (2025-01-03T19:28:53Z)
• Transferable Post-training via Inverse Value Learning [83.75002867411263]
  We propose modeling changes at the logits level during post-training using a separate neural network (i.e., the value network) After training this network on a small base model using demonstrations, this network can be seamlessly integrated with other pre-trained models during inference. We demonstrate that the resulting value network has broad transferability across pre-trained models of different parameter sizes.
  arXiv  Detail & Related papers  (2024-10-28T13:48:43Z)
• 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)
• Distilling Monolingual and Crosslingual Word-in-Context Representations [18.87665111304974]
  We propose a method that distils representations of word meaning in context from a pre-trained language model in both monolingual and crosslingual settings. Our method does not require human-annotated corpora nor updates of the parameters of the pre-trained model. Our method learns to combine the outputs of different hidden layers of the pre-trained model using self-attention.
  arXiv  Detail & Related papers  (2024-09-13T11:10:16Z)
• Data Shapley in One Training Run [88.59484417202454]
  Data Shapley provides a principled framework for attributing data's contribution within machine learning contexts. Existing approaches require re-training models on different data subsets, which is computationally intensive. This paper introduces In-Run Data Shapley, which addresses these limitations by offering scalable data attribution for a target model of interest.
  arXiv  Detail & Related papers  (2024-06-16T17:09:24Z)
• Causal Estimation of Memorisation Profiles [58.20086589761273]
  Understanding memorisation in language models has practical and societal implications. Memorisation is the causal effect of training with an instance on the model's ability to predict that instance. This paper proposes a new, principled, and efficient method to estimate memorisation based on the difference-in-differences design from econometrics.
  arXiv  Detail & Related papers  (2024-06-06T17:59:09Z)
• The Fine Line: Navigating Large Language Model Pretraining with Down-streaming Capability Analysis [27.310894780313618]
  This paper undertakes a comprehensive comparison of model capabilities at various pretraining intermediate checkpoints. We confirm that specific downstream metrics exhibit similar training dynamics across models of different sizes. In addition to our core findings, we've reproduced Amber and OpenLLaMA, releasing their intermediate checkpoints.
  arXiv  Detail & Related papers  (2024-04-01T16:00:01Z)
• 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.<n>We propose nested use of the scaling laws of training steps, model sizes, and our data mixing law.<n>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)
• Beyond Simple Averaging: Improving NLP Ensemble Performance with Topological-Data-Analysis-Based Weighting [2.6862667248315386]
  In natural language processing, ensembles boost the performance of a method due to multiple large models available in open source.<n>We propose to estimate weights for ensembles of NLP models using not only knowledge of their individual performance but also their similarity to each other.
  arXiv  Detail & Related papers  (2024-02-22T00:04:21Z)
• Do Smaller Language Models Answer Contextualised Questions Through Memorisation Or Generalisation? [8.51696622847778]
  A distinction is often drawn between a model's ability to predict a label for an evaluation sample that is directly memorised from highly similar training samples. We propose a method of identifying evaluation samples for which it is very unlikely our model would have memorised the answers.
  arXiv  Detail & Related papers  (2023-11-21T04:06:08Z)
• Data Similarity is Not Enough to Explain Language Model Performance [6.364065652816667]
  Similarity measures correlate with language model performance. Similarity metrics are not correlated with accuracy or even each other. This suggests that the relationship between pretraining data and downstream tasks is more complex than often assumed.
  arXiv  Detail & Related papers  (2023-11-15T14:48:08Z)
• Self-Influence Guided Data Reweighting for Language Model Pre-training [46.57714637505164]
  Language Models (LMs) pre-trained with self-supervision on large text corpora have become the default starting point for developing models for various NLP tasks. All data samples in the corpus are treated with equal importance during LM pre-training. Due to varying levels of relevance and quality of data, equal importance to all the data samples may not be the optimal choice. We propose PRESENCE, a method for jointly reweighting samples by leveraging self-influence (SI) scores as an indicator of sample importance and pre-training.
  arXiv  Detail & Related papers  (2023-11-02T01:00:46Z)
• Learning Evaluation Models from Large Language Models for Sequence Generation [61.8421748792555]
  We propose a three-stage evaluation model training method that utilizes large language models to generate labeled data for model-based metric development.<n> Experimental results on the SummEval benchmark demonstrate that CSEM can effectively train an evaluation model without human-labeled data.
  arXiv  Detail & Related papers  (2023-08-08T16:41:16Z)
• Dataless Knowledge Fusion by Merging Weights of Language Models [51.8162883997512]
  Fine-tuning pre-trained language models has become the prevalent paradigm for building downstream NLP models. This creates a barrier to fusing knowledge across individual models to yield a better single model. We propose a dataless knowledge fusion method that merges models in their parameter space.
  arXiv  Detail & Related papers  (2022-12-19T20:46:43Z)
• Uncertainty Estimation for Language Reward Models [5.33024001730262]
  Language models can learn a range of capabilities from unsupervised training on text corpora. It is often easier for humans to choose between options than to provide labeled data, and prior work has achieved state-of-the-art performance by training a reward model from such preference comparisons. We seek to address these problems via uncertainty estimation, which can improve sample efficiency and robustness using active learning and risk-averse reinforcement learning.
  arXiv  Detail & Related papers  (2022-03-14T20:13:21Z)
• Tracing Origins: Coref-aware Machine Reading Comprehension [43.352833140317486]
  We imitated the human's reading process in connecting the anaphoric expressions and leverage the coreference information to enhance the word embeddings from the pre-trained model. We demonstrated that the explicit incorporation of the coreference information in fine-tuning stage performed better than the incorporation of the coreference information in training a pre-trained language models.
  arXiv  Detail & Related papers  (2021-10-15T09:28:35Z)
• Instance-Based Neural Dependency Parsing [56.63500180843504]
  We develop neural models that possess an interpretable inference process for dependency parsing. Our models adopt instance-based inference, where dependency edges are extracted and labeled by comparing them to edges in a training set.
  arXiv  Detail & Related papers  (2021-09-28T05:30:52Z)
• Unsupervised Paraphrasing with Pretrained Language Models [85.03373221588707]
  We propose a training pipeline that enables pre-trained language models to generate high-quality paraphrases in an unsupervised setting. Our recipe consists of task-adaptation, self-supervision, and a novel decoding algorithm named Dynamic Blocking. We show with automatic and human evaluations that our approach achieves state-of-the-art performance on both the Quora Question Pair and the ParaNMT datasets.
  arXiv  Detail & Related papers  (2020-10-24T11:55:28Z)
• Adaptive Discrete Smoothing for High-Dimensional and Nonlinear Panel Data [4.550919471480445]
  We develop a data-driven smoothing technique for high-dimensional and non-linear panel data models. The weights are determined by a data-driven way and depend on the similarity between the corresponding functions. We conduct a simulation study which shows that the prediction can be greatly improved by using our estimator.
  arXiv  Detail & Related papers  (2019-12-30T09:50:58Z)

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.