ACL Findings2024
Can Large Language Models Follow Concept Annotation Guidelines? A Case Study on Scientific and Financial Domains
Abstract
Although large language models (LLMs) exhibit remarkable capacity to leverage in-context demonstrations, it is still unclear to what extent they can learn new facts or concept definitions via prompts. To address this question, we examine the capacity of instruction-tuned LLMs to follow in-context concept annotation guidelines for zero-shot sentence labeling tasks. We design guidelines that present different types of factual and counterfactual concept definitions, which are used as prompts for zero-shot sentence classification tasks. Our results show that although concept definitions consistently help in task performance, only the larger models (with 70B parameters or more) have limited ability to work under counterfactual contexts. Importantly, only proprietary models such as GPT-3.5 can recognize nonsensical guidelines, which we hypothesize is due to more sophisticated alignment methods. Finally, we find that FALCON-180B-CHAT is outperformed by LLAMA-2-70B-CHAT is most cases, which indicates that increasing model scale does not guarantee better adherence to guidelines. Altogether, our simple evaluation method reveals significant gaps in concept understanding between the most capable open-source language models and the leading proprietary APIs. 1
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
2 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Locating and Editing Factual Associations in GPT cites Language Models are Few-Shot Learners
- Large language models are few-shot clinical information extractors cites Language Models are Few-Shot Learners
- Large language models are few-shot clinical information extractors cites Training language models to follow instructions with human feedback
- An Explanation of In-context Learning as Implicit Bayesian Inference cites Language Models are Few-Shot Learners
- Training language models to follow instructions with human feedback cites Language Models are Few-Shot Learners
Cited by top-tier papers2
Builds on10
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Training language models to follow instructions with human feedbackLong Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida et al.NeurIPS 2022 · 24,707 citations
- Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
- An Explanation of In-context Learning as Implicit Bayesian InferenceSang Michael Xie, Aditi Raghunathan, Percy Liang, Tengyu MaICLR 2022 · 1,030 citations
- Large language models are few-shot clinical information extractorsMonica Agrawal, Stefan Hegselmann, Hunter Lang, Yoon Kim et al.EMNLP 2022 · 285 citations
Related papers
- GuideNER: Annotation Guidelines Are Better than Examples for In-Context Named Entity RecognitionShizhou Huang, Bo Xu, Yang Yu, Changqun Li et al.AAAI 2025 · 1 citation
- Definitions Matter: Guiding GPT for Multi-label ClassificationYouri Peskine, Damir Korencic, Ivan Grubisic, Paolo Papotti et al.EMNLP Findings 2023 · 10 citations
- Zero-Shot Multi-Label Topic Inference with Sentence Encoders and LLMsSouvika Sarkar, Dongji Feng, Shubhra Kanti Karmaker SantuEMNLP 2023 · 5 citations
- On the Limits of LLM Adaptability: Impact of LLM Pre-Training on Annotation Task PerformanceEtienne Casanova, Rafal Kocielnik, R. Michael AlvarezICML 2026
- Agent Instructs Large Language Models to be General Zero-Shot ReasonersNicholas Crispino, Kyle Montgomery, Fankun Zeng, Dawn Song et al.ICML 2024 · 41 citations
