Scaling Sparse and Dense Retrieval in Decoder-Only LLMs

• URL: http://arxiv.org/abs/2502.15526v1
• Date: Fri, 21 Feb 2025 15:28:26 GMT
• Title: Scaling Sparse and Dense Retrieval in Decoder-Only LLMs
• Authors: Hansi Zeng, Julian Killingback, Hamed Zamani,
• Abstract summary: Scaling large language models (LLMs) has shown great potential for improving retrieval model performance.<n>Previous studies have mainly focused on dense retrieval trained with contrastive loss (CL)<n>Sparse retrieval models consistently outperform dense retrieval across both in-domain (MSMARCO, TREC DL) and out-of-domain (BEIR) benchmarks.
• Score: 20.173669986209024
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Scaling large language models (LLMs) has shown great potential for improving retrieval model performance; however, previous studies have mainly focused on dense retrieval trained with contrastive loss (CL), neglecting the scaling behavior of other retrieval paradigms and optimization techniques, such as sparse retrieval and knowledge distillation (KD). In this work, we conduct a systematic comparative study on how different retrieval paradigms (sparse vs. dense) and fine-tuning objectives (CL vs. KD vs. their combination) affect retrieval performance across different model scales. Using MSMARCO passages as the training dataset, decoder-only LLMs (Llama-3 series: 1B, 3B, 8B), and a fixed compute budget, we evaluate various training configurations on both in-domain (MSMARCO, TREC DL) and out-of-domain (BEIR) benchmarks. Our key findings reveal that: (1) Scaling behaviors emerge clearly only with CL, where larger models achieve significant performance gains, whereas KD-trained models show minimal improvement, performing similarly across the 1B, 3B, and 8B scales. (2) Sparse retrieval models consistently outperform dense retrieval across both in-domain (MSMARCO, TREC DL) and out-of-domain (BEIR) benchmarks, and they demonstrate greater robustness to imperfect supervised signals. (3) We successfully scale sparse retrieval models with the combination of CL and KD losses at 8B scale, achieving state-of-the-art (SOTA) results in all evaluation sets.

Related papers

• CSPLADE: Learned Sparse Retrieval with Causal Language Models [12.930248566238243]
  We identify two challenges in training large language models (LLM) for Learned sparse retrieval (LSR) We propose two corresponding techniques: (1) a lightweight adaptation training phase to eliminate training instability; (2) two model variants to enable bidirectional information. With these techniques, we are able to train LSR models with 8B scale LLM, and achieve competitive retrieval performance with reduced index size.
  arXiv  Detail & Related papers  (2025-04-15T02:31:34Z)
• Exploring Training and Inference Scaling Laws in Generative Retrieval [50.82554729023865]
  We investigate how model size, training data scale, and inference-time compute jointly influence generative retrieval performance. Our experiments show that n-gram-based methods demonstrate strong alignment with both training and inference scaling laws. We find that LLaMA models consistently outperform T5 models, suggesting a particular advantage for larger decoder-only models in generative retrieval.
  arXiv  Detail & Related papers  (2025-03-24T17:59:03Z)
• Dynamic Loss-Based Sample Reweighting for Improved Large Language Model Pretraining [55.262510814326035]
  Existing reweighting strategies primarily focus on group-level data importance.<n>We introduce novel algorithms for dynamic, instance-level data reweighting.<n>Our framework allows us to devise reweighting strategies deprioritizing redundant or uninformative data.
  arXiv  Detail & Related papers  (2025-02-10T17:57:15Z)
• USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation [24.90512145836643]
  We introduce a Unified Skeleton-based Dense Representation Learning framework based on feature decorrelation.<n>We show that our approach significantly outperforms the current state-of-the-art (SOTA) approaches.
  arXiv  Detail & Related papers  (2024-12-12T12:20:27Z)
• What Do Learning Dynamics Reveal About Generalization in LLM Reasoning? [83.83230167222852]
  We find that a model's generalization behavior can be effectively characterized by a training metric we call pre-memorization train accuracy. By connecting a model's learning behavior to its generalization, pre-memorization train accuracy can guide targeted improvements to training strategies.
  arXiv  Detail & Related papers  (2024-11-12T09:52:40Z)
• In-context Learning and Gradient Descent Revisited [3.085927389171139]
  We show that even untrained models achieve comparable ICL-GD similarity scores despite not exhibiting ICL. Next, we explore a major discrepancy in the flow of information throughout the model between ICL and GD, which we term Layer Causality. We propose a simple GD-based optimization procedure that respects layer causality, and show it improves similarity scores significantly.
  arXiv  Detail & Related papers  (2023-11-13T21:42:38Z)
• The Languini Kitchen: Enabling Language Modelling Research at Different Scales of Compute [66.84421705029624]
  We introduce an experimental protocol that enables model comparisons based on equivalent compute, measured in accelerator hours. We pre-process an existing large, diverse, and high-quality dataset of books that surpasses existing academic benchmarks in quality, diversity, and document length. This work also provides two baseline models: a feed-forward model derived from the GPT-2 architecture and a recurrent model in the form of a novel LSTM with ten-fold throughput.
  arXiv  Detail & Related papers  (2023-09-20T10:31:17Z)
• Class Anchor Margin Loss for Content-Based Image Retrieval [97.81742911657497]
  We propose a novel repeller-attractor loss that falls in the metric learning paradigm, yet directly optimize for the L2 metric without the need of generating pairs. We evaluate the proposed objective in the context of few-shot and full-set training on the CBIR task, by using both convolutional and transformer architectures.
  arXiv  Detail & Related papers  (2023-06-01T12:53:10Z)
• Foundational Models for Continual Learning: An Empirical Study of Latent Replay [17.322679682451597]
  We study the efficacy of pre-trained vision models as a foundation for downstream continual learning scenarios. We compare efficacy of various pre-trained models in large-scale benchmarking scenarios with a vanilla replay setting applied in the latent and in the raw-data space.
  arXiv  Detail & Related papers  (2022-04-30T19:11:37Z)
• Self-Supervised Class Incremental Learning [51.62542103481908]
  Existing Class Incremental Learning (CIL) methods are based on a supervised classification framework sensitive to data labels. When updating them based on the new class data, they suffer from catastrophic forgetting: the model cannot discern old class data clearly from the new. In this paper, we explore the performance of Self-Supervised representation learning in Class Incremental Learning (SSCIL) for the first time.
  arXiv  Detail & Related papers  (2021-11-18T06:58:19Z)
• SPLADE v2: Sparse Lexical and Expansion Model for Information Retrieval [11.38022203865326]
  SPLADE model provides highly sparse representations and competitive results with respect to state-of-the-art dense and sparse approaches. We modify the pooling mechanism, benchmark a model solely based on document expansion, and introduce models trained with distillation. Overall, SPLADE is considerably improved with more than $9$% gains on NDCG@10 on TREC DL 2019, leading to state-of-the-art results on the BEIR benchmark.
  arXiv  Detail & Related papers  (2021-09-21T10:43:42Z)

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.