Are Pretrained Language Models Symbolic Reasoners Over Knowledge?

• URL: http://arxiv.org/abs/2006.10413v2
• Date: Sat, 10 Oct 2020 10:09:46 GMT
• Title: Are Pretrained Language Models Symbolic Reasoners Over Knowledge?
• Authors: Nora Kassner, Benno Krojer, Hinrich Sch\"utze
• Abstract summary: We investigate the two most important mechanisms: reasoning and memorization. For reasoning, we show that PLMs seem to learn to apply some symbolic reasoning rules correctly but struggle with others, including two-hop reasoning. For memorization, we identify schema conformity (facts systematically supported by other facts) and frequency as key factors for its success.
• Score: 5.480912891689259
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: How can pretrained language models (PLMs) learn factual knowledge from the training set? We investigate the two most important mechanisms: reasoning and memorization. Prior work has attempted to quantify the number of facts PLMs learn, but we present, using synthetic data, the first study that investigates the causal relation between facts present in training and facts learned by the PLM. For reasoning, we show that PLMs seem to learn to apply some symbolic reasoning rules correctly but struggle with others, including two-hop reasoning. Further analysis suggests that even the application of learned reasoning rules is flawed. For memorization, we identify schema conformity (facts systematically supported by other facts) and frequency as key factors for its success.

Related papers

• Understanding Fact Recall in Language Models: Why Two-Stage Training Encourages Memorization but Mixed Training Teaches Knowledge [21.798525556259378]
  We investigate how training strategies affect how model parameters are shaped during training and how these differences relate to their ability to recall facts.<n>Our analysis on synthetic fact recall datasets with the Llama-3.2B and Pythia-2.8B models reveals that mixed training encouraging a larger and more centralized set of shared parameters.
  arXiv  Detail & Related papers  (2025-05-22T03:34:29Z)
• On Memorization of Large Language Models in Logical Reasoning [70.94164038947078]
  Large language models (LLMs) achieve good performance on challenging reasoning benchmarks, yet could also make basic reasoning mistakes. One hypothesis is that the increasingly high and nearly saturated performance could be due to the memorization of similar problems. We show that fine-tuning leads to heavy memorization, but it also consistently improves generalization performance.
  arXiv  Detail & Related papers  (2024-10-30T15:31:54Z)
• Reasoning with Large Language Models, a Survey [2.831296564800826]
  This paper reviews the rapidly expanding field of prompt-based reasoning with LLMs. Our taxonomy identifies different ways to generate, evaluate, and control multi-step reasoning. We find that self-improvement, self-reflection, and some meta abilities of the reasoning processes are possible through the judicious use of prompts.
  arXiv  Detail & Related papers  (2024-07-16T08:49:35Z)
• Chain-of-Knowledge: Integrating Knowledge Reasoning into Large Language Models by Learning from Knowledge Graphs [55.317267269115845]
  Chain-of-Knowledge (CoK) is a comprehensive framework for knowledge reasoning. CoK includes methodologies for both dataset construction and model learning. We conduct extensive experiments with KnowReason.
  arXiv  Detail & Related papers  (2024-06-30T10:49:32Z)
• What Matters in Memorizing and Recalling Facts? Multifaceted Benchmarks for Knowledge Probing in Language Models [15.057992220389604]
  Language models often struggle with handling factual knowledge, exhibiting factual hallucination issue. We introduce a knowledge probing benchmark, BELIEF(ICL), to evaluate the knowledge recall ability of both encoder- and decoder-based pre-trained language models. We semi-automatically create MyriadLAMA, which has massively diverse prompts.
  arXiv  Detail & Related papers  (2024-06-18T05:11:35Z)
• LogicBench: Towards Systematic Evaluation of Logical Reasoning Ability of Large Language Models [52.03659714625452]
  Recently developed large language models (LLMs) have been shown to perform remarkably well on a wide range of language understanding tasks. But, can they really "reason" over the natural language? This question has been receiving significant research attention and many reasoning skills such as commonsense, numerical, and qualitative have been studied.
  arXiv  Detail & Related papers  (2024-04-23T21:08:49Z)
• Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models [2.3981254787726067]
  Pre-trained Language Models (PLMs) are trained on vast unlabeled data, rich in world knowledge. This has sparked the interest of the community in quantifying the amount of factual knowledge present in PLMs. In this work, we survey methods and datasets that are used to probe PLMs for factual knowledge.
  arXiv  Detail & Related papers  (2023-10-25T11:57:13Z)
• Large Language Models can Learn Rules [106.40747309894236]
  We present Hypotheses-to-Theories (HtT), a framework that learns a rule library for reasoning with large language models (LLMs) Experiments on relational reasoning, numerical reasoning and concept learning problems show that HtT improves existing prompting methods. The learned rules are also transferable to different models and to different forms of the same problem.
  arXiv  Detail & Related papers  (2023-10-10T23:07:01Z)
• Do PLMs Know and Understand Ontological Knowledge? [72.48752398867651]
  Ontological knowledge comprises classes and properties and their relationships. It is significant to explore whether Pretrained Language Models (PLMs) know and understand such knowledge. Our results show that PLMs can memorize certain ontological knowledge and utilize implicit knowledge in reasoning.
  arXiv  Detail & Related papers  (2023-09-12T03:20:50Z)
• Can Pretrained Language Models (Yet) Reason Deductively? [72.9103833294272]
  We conduct a comprehensive evaluation of the learnable deductive (also known as explicit) reasoning capability of PLMs. Our main results suggest that PLMs cannot yet perform reliable deductive reasoning. We reach beyond (misleading) task performance, revealing that PLMs are still far from human-level reasoning capabilities.
  arXiv  Detail & Related papers  (2022-10-12T17:44:15Z)
• How Pre-trained Language Models Capture Factual Knowledge? A Causal-Inspired Analysis [43.86843444052966]
  We show how PLMs generate missing words by relying on effective clues or shortcut patterns. We check the words that have three typical associations with the missing words: knowledge-dependent, positionally close, and highly co-occurred. We conclude that the PLMs capture the factual knowledge ineffectively because of depending on the inadequate associations.
  arXiv  Detail & Related papers  (2022-03-31T02:01:26Z)
• KMIR: A Benchmark for Evaluating Knowledge Memorization, Identification and Reasoning Abilities of Language Models [28.82149012250609]
  We propose a benchmark, named Knowledge Memorization, Identification, and Reasoning test (KMIR) KMIR covers 3 types of knowledge, including general knowledge, domain-specific knowledge, and commonsense, and provides 184,348 well-designed questions. Preliminary experiments with various representative pre-training language models on KMIR reveal many interesting phenomenons.
  arXiv  Detail & Related papers  (2022-02-28T03:52:57Z)
• Do Language Embeddings Capture Scales? [54.1633257459927]
  We show that pretrained language models capture a significant amount of information about the scalar magnitudes of objects. We identify contextual information in pre-training and numeracy as two key factors affecting their performance.
  arXiv  Detail & Related papers  (2020-10-11T21:11:09Z)

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.