Prompt Embedding Probes (PEP): Hallucination Detection in LLMs from Hidden States

• URL: http://arxiv.org/abs/2608.08024v1
• Date: Sat, 08 Aug 2026 09:14:47 GMT
• Title: Prompt Embedding Probes (PEP): Hallucination Detection in LLMs from Hidden States
• Abstract summary: Large language models (LLMs) can generate fluent and useful responses but remain prone to hallucinations.<n>We introduce Prompt Embedding Probes (PEP), a white-box method for answer-level hallucination detection from the hidden states of a frozen LLM.<n> PEP improves hidden-state-based detection over standard linear probes in the main in-distribution setting.
• Score: 4.37919186746582
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) can generate fluent and useful responses but remain prone to hallucinations. We introduce Prompt Embedding Probes (PEP), a white-box method for answer-level hallucination detection from the hidden states of a frozen LLM. PEP extends standard linear probes by augmenting the input with a small number of learnable prompt embeddings. We evaluate PEP on TriviaQA, GSM8K, and MedQA using Qwen3 models at multiple scales. PEP improves hidden-state-based detection over standard linear probes in the main in-distribution setting. We further evaluate PEP for pre-generation prediction, cross-model transfer, and out-of-distribution generalization. PEP remains effective in the pre-generation and cross-model settings, whereas robust cross-dataset transfer remains difficult. These results show that prompt-based adaptation can strengthen hidden-state probing while keeping the backbone frozen and adding only a small number of trainable parameters.

Related papers

• AMPBench-MT: A Homology-Controlled Benchmark for Antimicrobial Peptide Potency, Spectrum, and Safety Prediction [3.534208235875249]
  AMP and peptide benchmarks do not jointly place AMP recognition, species-conditioned potency, spectrum, safety-facing proxy endpoints, and cross-endpoint behavior.<n> AMPBench-MT standardizes canonical peptide records and organizes them into binary recognition, species-conditioned pMIC regression, and endpoint-specific potency and safety-facing readouts.
  arXiv  Detail & Related papers  (2026-07-28T10:01:30Z)
• Measuring the Impact of Lexical Training Data Coverage on Hallucination Detection in Large Language Models [26.89705770151822]
  Hallucination in large language models (LLMs) is a fundamental challenge, particularly in open-domain question answering.<n>Prior work attempts to detect hallucination with model-internal signals such as token-level entropy or generation consistency.<n>We investigate whether data coverage itself can serve as a detection signal.
  arXiv  Detail & Related papers  (2025-11-22T06:59:55Z)
• Cross-Layer Attention Probing for Fine-Grained Hallucination Detection [6.83291363146574]
  We propose Cross-Layer Attention Probing (CLAP), a novel activation probing technique for hallucination detection.<n>Our empirical evaluations show that CLAP improves hallucination detection compared to baselines on both decoded responses and responses sampled at higher temperatures.<n>CLAP maintains high reliability even when applied out-of-distribution.
  arXiv  Detail & Related papers  (2025-09-04T14:37:34Z)
• Mitigating Hallucinations in Multimodal LLMs via Object-aware Preference Optimization [55.543583937522804]
  Multimodal Large Language Models (MLLMs) emerge as a unified interface to address a multitude of tasks.<n>Despite showcasing state-of-the-art results in many benchmarks, a long-standing issue is the tendency of MLLMs to hallucinate.<n>In this paper, we address the problem of hallucinations as an alignment problem, seeking to steer the MLLM so that it prefers generating content without hallucinations.
  arXiv  Detail & Related papers  (2025-08-27T18:02:04Z)
• ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs [50.18087419133284]
  hallucination detection methods leveraging hidden states predominantly focus on static and isolated representations.<n>We introduce a novel metric, the ICR Score, which quantifies the contribution of modules to the hidden states' update.<n>We propose a hallucination detection method, the ICR Probe, which captures the cross-layer evolution of hidden states.
  arXiv  Detail & Related papers  (2025-07-22T11:44:26Z)
• REAL Sampling: Boosting Factuality and Diversity of Open-Ended Generation via Asymptotic Entropy [93.8400683020273]
  Decoding methods for large language models (LLMs) usually struggle with the tradeoff between ensuring factuality and maintaining diversity. We propose REAL sampling, a decoding method that improved factuality and diversity over nucleus sampling.
  arXiv  Detail & Related papers  (2024-06-11T21:44:49Z)
• On Optimal Sampling for Learning SDF Using MLPs Equipped with Positional Encoding [76.78321723602636]
  We devise a tool to determine the appropriate sampling rate for learning an accurate neural implicit field without undesirable side effects.<n>It is observed that a PE-equipped has an intrinsic frequency much higher than the highest frequency component in the PE layer.<n>We empirically show in the setting of SDF fitting, this recommended sampling rate is sufficient to secure accurate fitting results.
  arXiv  Detail & Related papers  (2024-01-02T10:51:52Z)
• Robust Quantitative Susceptibility Mapping via Approximate Message Passing with Parameter Estimation [14.22930572798757]
  We propose a probabilistic Bayesian approach for quantitative susceptibility mapping (QSM) with built-in parameter estimation. On the simulated Sim2Snr1 dataset, AMP-PE achieved the lowest NRMSE, DFCM and the highest SSIM. On the in vivo datasets, AMP-PE is robust and successfully recovers the susceptibility maps using the estimated parameters.
  arXiv  Detail & Related papers  (2022-07-29T14:38:03Z)
• Transformers Can Do Bayesian Inference [56.99390658880008]
  We present Prior-Data Fitted Networks (PFNs) PFNs leverage in-context learning in large-scale machine learning techniques to approximate a large set of posteriors. We demonstrate that PFNs can near-perfectly mimic Gaussian processes and also enable efficient Bayesian inference for intractable problems.
  arXiv  Detail & Related papers  (2021-12-20T13:07:39Z)
• Detecting Hallucinated Content in Conditional Neural Sequence Generation [165.68948078624499]
  We propose a task to predict whether each token in the output sequence is hallucinated (not contained in the input) We also introduce a method for learning to detect hallucinations using pretrained language models fine tuned on synthetic data.
  arXiv  Detail & Related papers  (2020-11-05T00:18:53Z)

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.