White-box method ReXTrust achieves highest AUC (peak 93.0) on Gut-VLM across five VLMs, outperforming alternatives by statistically significant margins while black-box and some gray-box methods collapse on certain models.
hub Canonical reference
The Internal State of an LLM Knows When It's Lying
Canonical reference. 83% of citing Pith papers cite this work as background.
37
Pith papers citing it
Background
83% of classified citations
abstract
While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM's internal state can be used to reveal the truthfulness of statements. This includes both statements provided to the LLM, and statements that the LLM itself generates. Our approach is to train a classifier that outputs the probability that a statement is truthful, based on the hidden layer activations of the LLM as it reads or generates the statement. Experiments demonstrate that given a set of test sentences, of which half are true and half false, our trained classifier achieves an average of 71\% to 83\% accuracy labeling which sentences are true versus false, depending on the LLM base model. Furthermore, we explore the relationship between our classifier's performance and approaches based on the probability assigned to the sentence by the LLM. We show that while LLM-assigned sentence probability is related to sentence truthfulness, this probability is also dependent on sentence length and the frequencies of words in the sentence, resulting in our trained classifier providing a more reliable approach to detecting truthfulness, highlighting its potential to enhance the reliability of LLM-generated content and its practical applicability in real-world scenarios.
hub tools
citation-role summary
background 5
method 1
citation-polarity summary
representative citing papers
Hidden-state convergence at step 4 predicts behavioral consistency in LLM agents on QA tasks (r=-0.35 to -0.83), enabling AUROC 0.97 detection of inconsistent trajectories but not improving accuracy on harder benchmarks.
ActProbe is an action-space detector that uses temporal consistency error and action chunk magnitude from policy outputs, mapped via LSTM-MLP, to predict failures earlier than baselines across policies and real-robot tasks.
DICE formalizes multi-agent LLM coordination as discounted incomplete-information Markov games and introduces Heterogeneous Quantal Response Equilibrium (HQRE) to achieve unique stable equilibria with bounded regret, demonstrated via prompt-control and fine-tuning algorithms on eleven benchmarks.
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
LLM rerankers can internally predict ranking quality via self-consistency of sampled outputs, matching SOTA external QPP while direct confidence is overconfident; supervised token-efficient methods improve calibration.
BICR trains a lightweight probe on contrastive hidden states from real versus blind images to detect visual grounding in LVLM predictions, outperforming baselines on calibration and discrimination with fewer parameters.
Latent space probing on CogVideoX achieves 97.29% F1 for adult content detection on a new 11k-clip dataset with 4-6ms overhead.
Uncertainty and correctness in LLMs are encoded by distinct feature populations, with suppression of confounded features improving accuracy and reducing entropy.
Subword tokenization impairs phonological knowledge encoding in LMs, but an IPA-based fine-tuning method restores it with minimal impact on other capabilities.
RAGognizer adds a detection head to LLMs for joint training on generation and token-level hallucination detection, yielding SOTA detection and fewer hallucinations in RAG while preserving output quality.
NARCBench and five activation-probing methods detect multi-agent collusion with 0.73-1.00 AUROC across distribution shifts and steganographic tasks by aggregating per-agent signals.
A framework is proposed for solving for an AI system's beliefs and desires from its computational facts, with criteria for success tied to interpretability tests and emphasis on holistic attribution.
Grad Detect uses internal gradient patterns from one inference pass to predict LLM hallucinations and abstention, outperforming confidence and sampling baselines on Q&A benchmarks with most signal in the final five layers.
Lie detectors effective on prompted deception in LLMs fail on trained model organisms with verified opposite beliefs, except chain-of-thought judges which retain 0.82 balanced accuracy partly due to verification artifacts.
Steering LLM residual streams with random sparse vectors creates detectable self-recognition fingerprints that enable over 98% accurate attribution of generated text to specific models without degrading output quality.
The multi-source framework identifies a consistent relational deficit in LLMs, where they recognize domain concepts but fail to reproduce their relational structures when compared to expert encyclopedias across fields like sociology and philosophy.
SABER combines self-prior with multi-trace PK and CK reasoning representations to estimate reliability beliefs and drive trust-or-abstain decisions in knowledge-conflict RAG, improving accuracy over baselines.
DisAAD trains a 1%-sized proxy model via adversarial distillation to quantify uncertainty in black-box LLMs by aligning with their output distributions.
Activation steering reveals localized encoding for entities versus distributed encoding for abstract concepts in MLLMs, identifying depth as key for the latter and a perception-reasoning disconnect.
LLMs implement a second-order confidence architecture where the PANL activation encodes both error likelihood and the ability to correct it, beyond verbal confidence or log-probabilities.
Weak supervision signals can be distilled into LLM hidden states so that simple probes on internal activations detect hallucinations at inference without external tools.
RETINA-SAFE benchmark and ECRT two-stage triage improve hallucination risk detection in medical LLMs for retinal decisions by 0.15-0.19 balanced accuracy over baselines using internal representations and logit shifts.
Mechanistic experiments on Gemma 3 27B, Qwen 2.5 7B and Magistral Small 24B show verbal confidence is cached at post-answer positions from answer tokens and captures richer answer-quality information beyond token log-probabilities.
citing papers explorer
No citing papers match the current filters.