SHADE adaptively combines coverage and spectral signals to estimate semantic alphabet size from few LLM samples, yielding better performance than baselines in low-sample regimes for alphabet estimation and QA error detection.
Kernel language entropy: Fine-grained uncertainty quantification for llms from semantic similarities.ArXiv, abs/2405.20003
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
fields
cs.CL 5representative citing papers
Adapts multi-layer token-level Mahalanobis distance with supervised linear regression to yield improved uncertainty scores for LLM truthfulness tasks.
A regression model using attention features and recurrent uncertainty scores improves selective generation in LLMs over unsupervised and supervised baselines on ten datasets and three models.
Supervised fine-tuning degrades the correlation between confidence scores and output quality in language models, driven by factors like training distribution similarity rather than true quality.
citing papers explorer
-
Mind the Unseen Mass: Unmasking LLM Hallucinations via Soft-Hybrid Alphabet Estimation
SHADE adaptively combines coverage and spectral signals to estimate semantic alphabet size from few LLM samples, yielding better performance than baselines in low-sample regimes for alphabet estimation and QA error detection.
-
Token-Level Density-Based Uncertainty Quantification Methods for Eliciting Truthfulness of Large Language Models
Adapts multi-layer token-level Mahalanobis distance with supervised linear regression to yield improved uncertainty scores for LLM truthfulness tasks.
-
Unconditional Truthfulness: Learning Unconditional Uncertainty of Large Language Models
A regression model using attention features and recurrent uncertainty scores improves selective generation in LLMs over unsupervised and supervised baselines on ten datasets and three models.
-
Confident in a Confidence Score: Investigating the Sensitivity of Confidence Scores to Supervised Fine-Tuning
Supervised fine-tuning degrades the correlation between confidence scores and output quality in language models, driven by factors like training distribution similarity rather than true quality.
- REALISTA: Realistic Latent Adversarial Attacks that Elicit LLM Hallucinations