Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models

• URL: http://arxiv.org/abs/2411.12580v1
• Date: Tue, 19 Nov 2024 15:47:12 GMT
• Title: Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models
• Authors: Laura Ruis, Maximilian Mozes, Juhan Bae, Siddhartha Rao Kamalakara, Dwarak Talupuru, Acyr Locatelli, Robert Kirk, Tim Rocktäschel, Edward Grefenstette, Max Bartolo,
• Abstract summary: We study what kind of generalisation strategies Large Language Models employ when performing reasoning tasks. Our findings indicate that the approach to reasoning the models use is unlike retrieval, and more like a generalisable strategy.
• Score: 30.066436019078164
• License:
• Abstract: The capabilities and limitations of Large Language Models have been sketched out in great detail in recent years, providing an intriguing yet conflicting picture. On the one hand, LLMs demonstrate a general ability to solve problems. On the other hand, they show surprising reasoning gaps when compared to humans, casting doubt on the robustness of their generalisation strategies. The sheer volume of data used in the design of LLMs has precluded us from applying the method traditionally used to measure generalisation: train-test set separation. To overcome this, we study what kind of generalisation strategies LLMs employ when performing reasoning tasks by investigating the pretraining data they rely on. For two models of different sizes (7B and 35B) and 2.5B of their pretraining tokens, we identify what documents influence the model outputs for three simple mathematical reasoning tasks and contrast this to the data that are influential for answering factual questions. We find that, while the models rely on mostly distinct sets of data for each factual question, a document often has a similar influence across different reasoning questions within the same task, indicating the presence of procedural knowledge. We further find that the answers to factual questions often show up in the most influential data. However, for reasoning questions the answers usually do not show up as highly influential, nor do the answers to the intermediate reasoning steps. When we characterise the top ranked documents for the reasoning questions qualitatively, we confirm that the influential documents often contain procedural knowledge, like demonstrating how to obtain a solution using formulae or code. Our findings indicate that the approach to reasoning the models use is unlike retrieval, and more like a generalisable strategy that synthesises procedural knowledge from documents doing a similar form of reasoning.

Related papers

• Disentangling Memory and Reasoning Ability in Large Language Models [97.26827060106581]
  We propose a new inference paradigm that decomposes the complex inference process into two distinct and clear actions. Our experiment results show that this decomposition improves model performance and enhances the interpretability of the inference process.
  arXiv  Detail & Related papers  (2024-11-20T17:55:38Z)
• Formality is Favored: Unraveling the Learning Preferences of Large Language Models on Data with Conflicting Knowledge [55.65162959527848]
  Large language models have shown excellent performance on many knowledge-intensive tasks. However, pretraining data tends to contain misleading and even conflicting information. This study systematically analyze LLMs' learning preferences for data with conflicting knowledge.
  arXiv  Detail & Related papers  (2024-10-07T06:49:41Z)
• Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data [76.90128359866462]
  We introduce an extended concept of memorization, distributional memorization, which measures the correlation between the output probabilities and the pretraining data frequency. This study demonstrates that memorization plays a larger role in simpler, knowledge-intensive tasks, while generalization is the key for harder, reasoning-based tasks.
  arXiv  Detail & Related papers  (2024-07-20T21:24:40Z)
• Hierarchical Deconstruction of LLM Reasoning: A Graph-Based Framework for Analyzing Knowledge Utilization [30.349165483935682]
  How large language models (LLMs) use their knowledge for reasoning is not yet well understood. We develop the DepthQA dataset, deconstructing questions into three depths: (i) recalling conceptual knowledge, (ii) applying procedural knowledge, and (iii) analyzing strategic knowledge. Distinct patterns of discrepancies are observed across model capacity and possibility of training data memorization.
  arXiv  Detail & Related papers  (2024-06-27T19:29:36Z)
• Evaluating Consistency and Reasoning Capabilities of Large Language Models [0.0]
  Large Language Models (LLMs) are extensively used today across various sectors, including academia, research, business, and finance. Despite their widespread adoption, these models often produce incorrect and misleading information, exhibiting a tendency to hallucinate. This paper aims to evaluate and compare the consistency and reasoning capabilities of both public and proprietary LLMs.
  arXiv  Detail & Related papers  (2024-04-25T10:03:14Z)
• Can NLP Models Correctly Reason Over Contexts that Break the Common Assumptions? [14.991565484636745]
  We investigate the ability of NLP models to correctly reason over contexts that break the common assumptions. We show that while doing fairly well on contexts that follow the common assumptions, the models struggle to correctly reason over contexts that break those assumptions. Specifically, the performance gap is as high as 20% absolute points.
  arXiv  Detail & Related papers  (2023-05-20T05:20:37Z)
• Exploring Strategies for Generalizable Commonsense Reasoning with Pre-trained Models [62.28551903638434]
  We measure the impact of three different adaptation methods on the generalization and accuracy of models. Experiments with two models show that fine-tuning performs best, by learning both the content and the structure of the task, but suffers from overfitting and limited generalization to novel answers. We observe that alternative adaptation methods like prefix-tuning have comparable accuracy, but generalize better to unseen answers and are more robust to adversarial splits.
  arXiv  Detail & Related papers  (2021-09-07T03:13:06Z)
• Prompting Contrastive Explanations for Commonsense Reasoning Tasks [74.7346558082693]
  Large pretrained language models (PLMs) can achieve near-human performance on commonsense reasoning tasks. We show how to use these same models to generate human-interpretable evidence.
  arXiv  Detail & Related papers  (2021-06-12T17:06:13Z)
• Knowledge-driven Data Construction for Zero-shot Evaluation in Commonsense Question Answering [80.60605604261416]
  We propose a novel neuro-symbolic framework for zero-shot question answering across commonsense tasks. We vary the set of language models, training regimes, knowledge sources, and data generation strategies, and measure their impact across tasks. We show that, while an individual knowledge graph is better suited for specific tasks, a global knowledge graph brings consistent gains across different tasks.
  arXiv  Detail & Related papers  (2020-11-07T22:52:21Z)

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.