Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
139 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
46 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

On the Tip of the Tongue: Analyzing Conceptual Representation in Large Language Models with Reverse-Dictionary Probe (2402.14404v2)

Published 22 Feb 2024 in cs.CL and cs.AI

Abstract: Probing and enhancing LLMs' reasoning capacity remains a crucial open question. Here we re-purpose the reverse dictionary task as a case study to probe LLMs' capacity for conceptual inference. We use in-context learning to guide the models to generate the term for an object concept implied in a linguistic description. Models robustly achieve high accuracy in this task, and their representation space encodes information about object categories and fine-grained features. Further experiments suggest that the conceptual inference ability as probed by the reverse-dictionary task predicts model's general reasoning performance across multiple benchmarks, despite similar syntactic generalization behaviors across models. Explorative analyses suggest that prompting LLMs with description$\Rightarrow$word examples may induce generalization beyond surface-level differences in task construals and facilitate models on broader commonsense reasoning problems.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (70)
  1. The falcon series of open language models. arXiv preprint arXiv:2311.16867.
  2. Emily M. Bender and Alexander Koller. 2020. Climbing towards NLU: On meaning, form, and understanding in the age of data. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5185–5198, Online. Association for Computational Linguistics.
  3. Pythia: A suite for analyzing large language models across training and scaling. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 2397–2430. PMLR.
  4. Marcel Binz and Eric Schulz. 2023a. Turning large language models into cognitive models. arXiv preprint arXiv:2306.03917.
  5. Marcel Binz and Eric Schulz. 2023b. Using cognitive psychology to understand gpt-3. Proceedings of the National Academy of Sciences, 120(6):e2218523120.
  6. Piqa: Reasoning about physical commonsense in natural language. Proceedings of the AAAI Conference on Artificial Intelligence, 34(05):7432–7439.
  7. ProtoQA: A question answering dataset for prototypical common-sense reasoning. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1122–1136, Online. Association for Computational Linguistics.
  8. Sparks of artificial general intelligence: Early experiments with gpt-4. arXiv preprint arXiv:2303.12712.
  9. Polysemy and brevity versus frequency in language. Computer Speech & Language, 58:19–50.
  10. Tyler A. Chang and Benjamin K. Bergen. 2022. Word acquisition in neural language models. Transactions of the Association for Computational Linguistics, 10:1–16.
  11. BoolQ: Exploring the surprising difficulty of natural yes/no questions. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 2924–2936, Minneapolis, Minnesota. Association for Computational Linguistics.
  12. Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457.
  13. Jeff Da and Jungo Kasai. 2019. Cracking the contextual commonsense code: Understanding commonsense reasoning aptitude of deep contextual representations. In Proceedings of the First Workshop on Commonsense Inference in Natural Language Processing, pages 1–12, Hong Kong, China. Association for Computational Linguistics.
  14. Language models show human-like content effects on reasoning. arXiv preprint arXiv:2207.07051.
  15. The Centre for Speech, Language and the Brain (CSLB) concept property norms. Behavior Research Methods, 46(4):1119–1127.
  16. Christiane Fellbaum. 1998. WordNet: An electronic lexical database. MIT press.
  17. Do neural language representations learn physical commonsense? arXiv preprint arXiv:1908.02899.
  18. Michael C. Frank. 2023. Openly accessible LLMs can help us to understand human cognition. Nature Human Behaviour, 7(11):1825–1827.
  19. SyntaxGym: An online platform for targeted evaluation of language models. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 70–76, Online. Association for Computational Linguistics.
  20. Finding alignments between interpretable causal variables and distributed neural representations. arXiv preprint arXiv:2303.02536.
  21. Learning word vectors for 157 languages. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. European Language Resources Association (ELRA).
  22. Wes Gurnee and Max Tegmark. 2024. Language models represent space and time. In The Twelfth International Conference on Learning Representations.
  23. Human-like intuitive behavior and reasoning biases emerged in large language models but disappeared in ChatGPT. Nature Computational Science, 3(10):833–838.
  24. Inductive reasoning in humans and large language models. Cognitive Systems Research, 83:101155.
  25. Hannes Hansen and Martin N Hebart. 2022. Semantic features of object concepts generated with GPT-3. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 44.
  26. Large language models meet cognitive science: Llms as tools, models, and participants. In Proceedings of the annual meeting of the cognitive science society, volume 45.
  27. Things: A database of 1,854 object concepts and more than 26,000 naturalistic object images. PLOS ONE, 14(10):1–24.
  28. Revealing the multidimensional mental representations of natural objects underlying human similarity judgements. Nature Human Behaviour, 4(11):1173–1185.
  29. Learning to understand phrases by embedding the dictionary. Transactions of the Association for Computational Linguistics, 4:17–30.
  30. The curious case of neural text degeneration. In International Conference on Learning Representations.
  31. A systematic assessment of syntactic generalization in neural language models. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1725–1744, Online. Association for Computational Linguistics.
  32. Mistral 7b. arXiv preprint arXiv:2310.06825.
  33. Large language models struggle to learn long-tail knowledge. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 15696–15707. PMLR.
  34. CLAWS4: The tagging of the British National Corpus. In COLING 1994 Volume 1: The 15th International Conference on Computational Linguistics, Kyoto, Japan.
  35. Implicit representations of meaning in neural language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 1813–1827, Online. Association for Computational Linguistics.
  36. Emergent world representations: Exploring a sequence model trained on a synthetic task. In The Eleventh International Conference on Learning Representations.
  37. Textbooks are all you need ii: phi-1.5 technical report. arXiv preprint arXiv:2309.05463.
  38. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
  39. Charles Lovering and Ellie Pavlick. 2022. Unit testing for concepts in neural networks. Transactions of the Association for Computational Linguistics, 10:1193–1208.
  40. Predicting human similarity judgments using large language models. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 44.
  41. Embers of autoregression: Understanding large language models through the problem they are trained to solve. arXiv preprint arXiv:2309.13638.
  42. Can a suit of armor conduct electricity? a new dataset for open book question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2381–2391, Brussels, Belgium. Association for Computational Linguistics.
  43. Rethinking the role of demonstrations: What makes in-context learning work? In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 11048–11064, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
  44. A property induction framework for neural language models. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 44.
  45. COMPS: Conceptual minimal pair sentences for testing robust property knowledge and its inheritance in pre-trained language models. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2928–2949, Dubrovnik, Croatia. Association for Computational Linguistics.
  46. Melanie Mitchell and David C. Krakauer. 2023. The debate over understanding in ai’s large language models. Proceedings of the National Academy of Sciences, 120(13):e2215907120.
  47. Gregory Murphy. 2004. The big book of concepts. MIT press.
  48. Barbara Partee et al. 1984. Compositionality. Varieties of formal semantics, 3:281–311.
  49. Roma Patel and Ellie Pavlick. 2022. Mapping language models to grounded conceptual spaces. In International Conference on Learning Representations.
  50. The refinedweb dataset for falcon llm: outperforming curated corpora with web data, and web data only. arXiv preprint arXiv:2306.01116.
  51. Steven Piantadosi and Felix Hill. 2022. Meaning without reference in large language models. In NeurIPS 2022 Workshop on Neuro Causal and Symbolic AI (nCSI).
  52. Social IQa: Commonsense reasoning about social interactions. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 4463–4473, Hong Kong, China. Association for Computational Linguistics.
  53. Bushra Siddique and M. M. Sufyan Beg. 2023. Reverse Dictionary Formation: State of the Art and Future Directions. SN Computer Science, 4(2):168.
  54. Robyn Speer. 2022. rspeer/wordfreq: v3.0.
  55. Conceptual structure coheres in human cognition but not in large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 722–738, Singapore. Association for Computational Linguistics.
  56. CommonsenseQA: A question answering challenge targeting commonsense knowledge. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4149–4158, Minneapolis, Minnesota. Association for Computational Linguistics.
  57. MosaicML NLP Team. 2023. Introducing mpt-7b: A new standard for open-source, commercially usable llms.
  58. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971.
  59. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
  60. Attention is all you need. In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc.
  61. BLiMP: The benchmark of linguistic minimal pairs for English. Transactions of the Association for Computational Linguistics, 8:377–392.
  62. Emergent analogical reasoning in large language models. Nature Human Behaviour, 7(9):1526–1541.
  63. Emergent abilities of large language models. Transactions on Machine Learning Research.
  64. Huggingface’s transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771.
  65. Reasoning or reciting? exploring the capabilities and limitations of language models through counterfactual tasks. arXiv preprint arXiv:2307.02477.
  66. Causal interventions expose implicit situation models for commonsense language understanding. In Findings of the Association for Computational Linguistics: ACL 2023, pages 13265–13293, Toronto, Canada. Association for Computational Linguistics.
  67. BERT for monolingual and cross-lingual reverse dictionary. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 4329–4338, Online. Association for Computational Linguistics.
  68. HellaSwag: Can a machine really finish your sentence? In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4791–4800, Florence, Italy. Association for Computational Linguistics.
  69. Multi-channel reverse dictionary model. Proceedings of the AAAI Conference on Artificial Intelligence, 34(01):312–319.
  70. Revealing interpretable object representations from human behavior. In International Conference on Learning Representations.
Citations (2)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now