Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting

• URL: http://arxiv.org/abs/2310.11324v2
• Date: Mon, 1 Jul 2024 22:28:01 GMT
• Title: Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting
• Authors: Melanie Sclar, Yejin Choi, Yulia Tsvetkov, Alane Suhr,
• Abstract summary: Large language models (LLMs) are adopted as a fundamental component of language technologies. We find that several widely used open-source LLMs are extremely sensitive to subtle changes in prompt format in few-shot settings. We propose an algorithm that rapidly evaluates a sampled set of plausible prompt formats for a given task, and reports the interval of expected performance without accessing model weights.
• Score: 68.19544657508509
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As large language models (LLMs) are adopted as a fundamental component of language technologies, it is crucial to accurately characterize their performance. Because choices in prompt design can strongly influence model behavior, this design process is critical in effectively using any modern pre-trained generative language model. In this work, we focus on LLM sensitivity to a quintessential class of meaning-preserving design choices: prompt formatting. We find that several widely used open-source LLMs are extremely sensitive to subtle changes in prompt formatting in few-shot settings, with performance differences of up to 76 accuracy points when evaluated using LLaMA-2-13B. Sensitivity remains even when increasing model size, the number of few-shot examples, or performing instruction tuning. Our analysis suggests that work evaluating LLMs with prompting-based methods would benefit from reporting a range of performance across plausible prompt formats, instead of the currently-standard practice of reporting performance on a single format. We also show that format performance only weakly correlates between models, which puts into question the methodological validity of comparing models with an arbitrarily chosen, fixed prompt format. To facilitate systematic analysis we propose FormatSpread, an algorithm that rapidly evaluates a sampled set of plausible prompt formats for a given task, and reports the interval of expected performance without accessing model weights. Furthermore, we present a suite of analyses that characterize the nature of this sensitivity, including exploring the influence of particular atomic perturbations and the internal representation of particular formats.

Related papers

• Benchmarking Prompt Sensitivity in Large Language Models [13.986971540998258]
  Large language Models (LLMs) are highly sensitive to variations in prompt formulation. This paper introduces a new task, Prompt Sensitivity Prediction, and a dataset designed to investigate the effects of slight prompt variations on LLM performance.
  arXiv  Detail & Related papers  (2025-02-09T23:01:03Z)
• Does Prompt Formatting Have Any Impact on LLM Performance? [10.869929764785464]
  This paper examines the impact of different prompt templates on Large Language Models (LLMs) performance. We evaluated their impact across tasks like natural language reasoning, code generation, and translation using OpenAI's GPT models. Experiments show that GPT-3.5-turbo's performance varies by up to 40% in a code translation task depending on the prompt template, while larger models like GPT-4 are more robust to these variations.
  arXiv  Detail & Related papers  (2024-11-15T19:26:38Z)
• Scalable Influence and Fact Tracing for Large Language Model Pretraining [14.598556308631018]
  Training data attribution (TDA) methods aim to attribute model outputs back to specific training examples. We refine existing gradient-based methods to work effectively at scale. We release our prompt set and model outputs, along with a web-based visualization tool to explore influential examples.
  arXiv  Detail & Related papers  (2024-10-22T20:39:21Z)
• LLMs Are Biased Towards Output Formats! Systematically Evaluating and Mitigating Output Format Bias of LLMs [69.40865293066885]
  We present the first systematic evaluation examining format bias in performance of large language models (LLMs) We present our empirical format bias evaluation spanning four commonly used categories -- multiple-choice question-answer, wrapping, list, and mapping.
  arXiv  Detail & Related papers  (2024-08-16T10:45:45Z)
• OLMES: A Standard for Language Model Evaluations [64.85905119836818]
  OLMES is a documented, practical, open standard for reproducible language model evaluations. It supports meaningful comparisons between smaller base models that require the unnatural "cloze" formulation of multiple-choice questions. OLMES includes well-considered, documented recommendations guided by results from existing literature as well as new experiments resolving open questions.
  arXiv  Detail & Related papers  (2024-06-12T17:37:09Z)
• Prompt Optimization with EASE? Efficient Ordering-aware Automated Selection of Exemplars [66.823588073584]
  Large language models (LLMs) have shown impressive capabilities in real-world applications. The quality of these exemplars in the prompt greatly impacts performance. Existing methods fail to adequately account for the impact of exemplar ordering on the performance.
  arXiv  Detail & Related papers  (2024-05-25T08:23:05Z)
• Experimental Design for Active Transductive Inference in Large Language Models [18.2671641610825]
  We use active learning for adaptive prompt design and call it Active In-context Prompt Design (AIPD) We design the LLM prompt by adaptively choosing few-shot examples from a training set to optimize performance on a test set. We propose two algorithms, GO and SAL, which differ in how the few-shot examples are chosen.
  arXiv  Detail & Related papers  (2024-04-12T23:27:46Z)
• A Thorough Examination of Decoding Methods in the Era of LLMs [72.65956436513241]
  Decoding methods play an indispensable role in converting language models from next-token predictors into practical task solvers. This paper provides a comprehensive and multifaceted analysis of various decoding methods within the context of large language models. Our findings reveal that decoding method performance is notably task-dependent and influenced by factors such as alignment, model size, and quantization.
  arXiv  Detail & Related papers  (2024-02-10T11:14:53Z)
• LLM-augmented Preference Learning from Natural Language [19.700169351688768]
  Large Language Models (LLMs) are equipped to deal with larger context lengths. LLMs can consistently outperform the SotA when the target text is large. Few-shot learning yields better performance than zero-shot learning.
  arXiv  Detail & Related papers  (2023-10-12T17:17:27Z)
• MetricPrompt: Prompting Model as a Relevance Metric for Few-shot Text Classification [65.51149771074944]
  MetricPrompt eases verbalizer design difficulty by reformulating few-shot text classification task into text pair relevance estimation task. We conduct experiments on three widely used text classification datasets across four few-shot settings. Results show that MetricPrompt outperforms manual verbalizer and other automatic verbalizer design methods across all few-shot settings.
  arXiv  Detail & Related papers  (2023-06-15T06:51:35Z)

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.