Pre-training vs. Fine-tuning: A Reproducibility Study on Dense Retrieval Knowledge Acquisition

• URL: http://arxiv.org/abs/2505.07166v1
• Date: Mon, 12 May 2025 01:24:00 GMT
• Title: Pre-training vs. Fine-tuning: A Reproducibility Study on Dense Retrieval Knowledge Acquisition
• Authors: Zheng Yao, Shuai Wang, Guido Zuccon,
• Abstract summary: Dense retrievers utilize pre-trained backbone language models (e.g., BERT, LLaMA) that are fine-tuned via contrastive learning to perform the task of encoding text into sense representations.<n>Recent research has questioned the role of fine-tuning vs. that of pre-training within dense retrievers.<n>Our study confirms that in DPR tuning, pre-trained knowledge underpins retrieval performance, with fine-tuning primarily adjusting neuron activation rather than reorganizing knowledge.
• Score: 28.48078856765935
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Dense retrievers utilize pre-trained backbone language models (e.g., BERT, LLaMA) that are fine-tuned via contrastive learning to perform the task of encoding text into sense representations that can be then compared via a shallow similarity operation, e.g. inner product. Recent research has questioned the role of fine-tuning vs. that of pre-training within dense retrievers, specifically arguing that retrieval knowledge is primarily gained during pre-training, meaning knowledge not acquired during pre-training cannot be sub-sequentially acquired via fine-tuning. We revisit this idea here as the claim was only studied in the context of a BERT-based encoder using DPR as representative dense retriever. We extend the previous analysis by testing other representation approaches (comparing the use of CLS tokens with that of mean pooling), backbone architectures (encoder-only BERT vs. decoder-only LLaMA), and additional datasets (MSMARCO in addition to Natural Questions). Our study confirms that in DPR tuning, pre-trained knowledge underpins retrieval performance, with fine-tuning primarily adjusting neuron activation rather than reorganizing knowledge. However, this pattern does not hold universally, such as in mean-pooled (Contriever) and decoder-based (LLaMA) models. We ensure full reproducibility and make our implementation publicly available at https://github.com/ielab/DenseRetriever-Knowledge-Acquisition.

Related papers

• Logic-Oriented Retriever Enhancement via Contrastive Learning [15.39205142672531]
  LORE (Logic ORiented Retriever Enhancement) introduces fine-grained contrastive learning to activate latent logical analysis capacity.<n>LORE requires no external upervision, resources, or pre-retrieval analysis, remains index-compatible, and consistently improves retrieval utility and downstream generation.
  arXiv  Detail & Related papers  (2026-02-01T09:30:04Z)
• Learning Mechanism Underlying NLP Pre-Training and Fine-Tuning [0.0]
  Twofold goals are examined; to understand the mechanism underlying successful pre-training and to determine the interplay between the pre-training accuracy and the fine-tuning of classification tasks.<n>Results were based on the BERT-6 architecture pre-trained on the Wikipedia dataset and fine-tuned on the FewRel and DBpedia classification tasks.
  arXiv  Detail & Related papers  (2025-09-03T15:32:50Z)
• Adapting Pretrained Language Models for Citation Classification via Self-Supervised Contrastive Learning [13.725832389453911]
  Citation classification is pivotal for scholarly analysis.<n>Previous works suggest fine-tuning pretrained language models (PLMs) on citation classification.<n>We present a novel framework, Citss, that adapts the PLMs to overcome these challenges.
  arXiv  Detail & Related papers  (2025-05-20T15:05:27Z)
• Log-Augmented Generation: Scaling Test-Time Reasoning with Reusable Computation [80.69067017594709]
  Large language models (LLMs) and their agentic counterparts struggle to retain reasoning from previous tasks.<n>We propose a novel framework, log-augmented generation (LAG) that directly reuses prior computation and reasoning from past logs at test time.<n>Our method significantly outperforms standard agentic systems that do not utilize logs.
  arXiv  Detail & Related papers  (2025-05-20T14:14:38Z)
• Dense Passage Retrieval: Is it Retrieving? [1.9797215742507548]
  We explore DPR-trained models mechanistically by using a combination of probing, layer activation analysis, and model editing. Our experiments show that DPR training decentralizes how knowledge is stored in the network, creating multiple access pathways to the same information. We also uncover a limitation in this training style: the internal knowledge of the pre-trained model bounds what the retrieval model can retrieve.
  arXiv  Detail & Related papers  (2024-02-16T19:28:52Z)
• Progressive Feature Adjustment for Semi-supervised Learning from Pretrained Models [39.42802115580677]
  Semi-supervised learning (SSL) can leverage both labeled and unlabeled data to build a predictive model. Recent literature suggests that naively applying state-of-the-art SSL with a pretrained model fails to unleash the full potential of training data. We propose to use pseudo-labels from the unlabelled data to update the feature extractor that is less sensitive to incorrect labels.
  arXiv  Detail & Related papers  (2023-09-09T01:57:14Z)
• Unsupervised Dense Retrieval with Relevance-Aware Contrastive Pre-Training [81.3781338418574]
  We propose relevance-aware contrastive learning. We consistently improve the SOTA unsupervised Contriever model on the BEIR and open-domain QA retrieval benchmarks. Our method can not only beat BM25 after further pre-training on the target corpus but also serves as a good few-shot learner.
  arXiv  Detail & Related papers  (2023-06-05T18:20:27Z)
• Noise-Robust Dense Retrieval via Contrastive Alignment Post Training [89.29256833403167]
  Contrastive Alignment POst Training (CAPOT) is a highly efficient finetuning method that improves model robustness without requiring index regeneration. CAPOT enables robust retrieval by freezing the document encoder while the query encoder learns to align noisy queries with their unaltered root. We evaluate CAPOT noisy variants of MSMARCO, Natural Questions, and Trivia QA passage retrieval, finding CAPOT has a similar impact as data augmentation with none of its overhead.
  arXiv  Detail & Related papers  (2023-04-06T22:16:53Z)
• Can BERT Refrain from Forgetting on Sequential Tasks? A Probing Study [68.75670223005716]
  We find that pre-trained language models like BERT have a potential ability to learn sequentially, even without any sparse memory replay. Our experiments reveal that BERT can actually generate high quality representations for previously learned tasks in a long term, under extremely sparse replay or even no replay.
  arXiv  Detail & Related papers  (2023-03-02T09:03:43Z)
• Great Truths are Always Simple: A Rather Simple Knowledge Encoder for Enhancing the Commonsense Reasoning Capacity of Pre-Trained Models [89.98762327725112]
  Commonsense reasoning in natural language is a desired ability of artificial intelligent systems. For solving complex commonsense reasoning tasks, a typical solution is to enhance pre-trained language models(PTMs) with a knowledge-aware graph neural network(GNN) encoder. Despite the effectiveness, these approaches are built on heavy architectures, and can't clearly explain how external knowledge resources improve the reasoning capacity of PTMs.
  arXiv  Detail & Related papers  (2022-05-04T01:27:36Z)
• On the Transferability of Pre-trained Language Models: A Study from Artificial Datasets [74.11825654535895]
  Pre-training language models (LMs) on large-scale unlabeled text data makes the model much easier to achieve exceptional downstream performance. We study what specific traits in the pre-training data, other than the semantics, make a pre-trained LM superior to their counterparts trained from scratch on downstream tasks.
  arXiv  Detail & Related papers  (2021-09-08T10:39:57Z)
• Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks [133.93803565077337]
  retrieval-augmented generation models combine pre-trained parametric and non-parametric memory for language generation. We show that RAG models generate more specific, diverse and factual language than a state-of-the-art parametric-only seq2seq baseline.
  arXiv  Detail & Related papers  (2020-05-22T21:34:34Z)
• REALM: Retrieval-Augmented Language Model Pre-Training [37.3178586179607]
  We augment language model pre-training with a latent knowledge retriever, which allows the model to retrieve and attend over documents from a large corpus such as Wikipedia. For the first time, we show how to pre-train such a knowledge retriever in an unsupervised manner. We demonstrate the effectiveness of Retrieval-Augmented Language Model pre-training (REALM) by fine-tuning on the challenging task of Open-domain Question Answering (Open-QA)
  arXiv  Detail & Related papers  (2020-02-10T18:40:59Z)

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.