Beyond Confidence: The Rhythms of Reasoning in Generative Models
This addresses the issue of unreliable LLM predictions for users in natural language processing by providing a new stability metric, though it is incremental as it builds on existing analysis methods.
The paper tackles the problem of large language models (LLMs) being sensitive to input variations, which undermines reliability, by introducing the Token Constraint Bound (δ_TCB), a metric that quantifies internal state robustness to perturbations, showing it correlates with prompt engineering and reveals instabilities missed by conventional metrics like perplexity.
Large Language Models (LLMs) exhibit impressive capabilities yet suffer from sensitivity to slight input context variations, hampering reliability. Conventional metrics like accuracy and perplexity fail to assess local prediction robustness, as normalized output probabilities can obscure the underlying resilience of an LLM's internal state to perturbations. We introduce the Token Constraint Bound ($δ_{\mathrm{TCB}}$), a novel metric that quantifies the maximum internal state perturbation an LLM can withstand before its dominant next-token prediction significantly changes. Intrinsically linked to output embedding space geometry, $δ_{\mathrm{TCB}}$ provides insights into the stability of the model's internal predictive commitment. Our experiments show $δ_{\mathrm{TCB}}$ correlates with effective prompt engineering and uncovers critical prediction instabilities missed by perplexity during in-context learning and text generation. $δ_{\mathrm{TCB}}$ offers a principled, complementary approach to analyze and potentially improve the contextual stability of LLM predictions.