A two-stage framework that first reconstructs source text from a hidden activation and then evaluates the resulting text with a separate frozen question-answering model, yielding an explicit and inspectable intermediate readout is introduced, enabling compact cross-model inverters to surpass donor-matched question-conditioned verbalizers while improving both semantic recoverability and lexical fidelity.
Abstract
Activation verbalization methods such as Activation Oracle and Natural Language Autoencoders decode hidden representations of large language models into human-readable natural language. However, existing methods can produce incomplete or hallucinated descriptions, making their activation verbalizations difficult to trust and use reliably in practice. To this end, we introduce AVPO, a two-stage framework that first reconstructs source text from a hidden activation and then evaluates the resulting text with a separate frozen question-answering model, yielding an explicit and inspectable intermediate readout. We further optimize the inverter with direct preference optimization (DPO), using rewards that capture both semantic recoverability and lexical fidelity. Across six text families, AVPO improves gist- and detail-level information recovery over the strongest baseline by up to 17.1 and 9.3 percentage points, respectively. Crucially, the gains arise from preference optimization rather than fine-tuning on selected reconstructions alone, enabling compact cross-model inverters to surpass donor-matched question-conditioned verbalizers while improving both semantic recoverability and lexical fidelity. Moreover, out-of-distribution case study shows that AVPO better recovers high-level semantics while fabricating fewer details.
Hallucination – the generation of fluent but
factually unsupported or fabricated content – remains
one of the most persistent obstacles to deploying Large
Language Models (LLMs) in high-stakes domains such
as healthcare, finance, and law. A wide range of
mitigation strategies has been proposed, spanning
several distinc...
Shweta Narayan Bamne, M. S. Shegokar· International Journal of Eng...· 0 citations
EviAnchor is proposed, a training-free and single-branch inference framework that preserves and reactivates visual evidence throughout generation in large vision-language models and demonstrates consistent improvements in visual grounding.
Sihang Jia, Shu-Liang Liu, Song-Bo Yang et al.· 0 citations
This work identifies a mechanistic precondition for hallucination - the failure to produce adequately grounded output - not the complete origin of naturally occurring hallucination - not the complete origin of naturally occurring hallucinations.
H. Sayed, Apoorv Singh, Kumar Aman et al.· 0 citations
It is demonstrated that LLMs struggle to handle negation effectively, applying affirmative knowledge inappropriately, leading to hallucinations that contradict the negated input, compromising the logical consistency and factual accuracy of LLM outputs.
Jaehyung Seo, Hyeonseok Moon, Heu-Jeoung Lim· ACM Transactions on Knowledg...· 0 citations
A multi-signal detection pipeline combining fine-tuned DeBERTa-v3 classification, Monte Carlo Dropout uncertainty quantification, and temperature-scaled calibration for response-level hallucination detection is presented.
Savor is introduced, a training framework that augments the output schema with token and answer confidence, optimises the policy with a Group Relative Policy Optimisation objective that penalises calibration error and poor abstention decisions, and uses the learned confidence at inference time to revisit visual evidenc...
Zian Ding, Zi-Lin Zhao, Ying-Jie He et al.· arXiv.org· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduNov 19, 2026
An MIT and Dana-Farber Institute-led team used large language models to examine nearly 4,000 studies and hundreds of code repositories, revealing gaps in the information needed to scrutinize and reuse published research.
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.