The paper presents a rate‑distortion framework for understanding factual hallucination in closed‑book question answering. It shows that even when a fact is observed, limited memory forces it to be stored approximately, leading to errors that can be bounded by a combination of compression distortion and missing coverage. The authors derive a theoretical lower bound on error and validate it with simulations and probes on modern language models.
By Xi Wang, Shijia Xu, Rongfeng Guo
The study measured the impact of a single training example on a GPT‑2 model by running 24 counterfactual experiments. 32 models were trained from scratch on OpenWebText, and at a specific training step a single batch row was replaced with a 194‑token passage under three conditions (fluent prose, fabricated subject, random characters) or left unchanged. Results showed that the passage was learned from one exposure and decayed, with measurable differences in cross‑entropy up to 50 steps after injection but no lasting effect at the final step.
By Zachary Speck, Asa Shepard
arXiv:2607. 14111v1 Announce Type: cross Abstract: Can small language models detect and report on perturbations their own internal activations?
By Ely Hahami, Ishaan Sinha, Lavik Jain
arXiv:2608. 11138v1 Announce Type: cross Abstract: We propose that a model's uncertainty about a token is reflected not only in the breadth of its output distribution but also in whether a confident prediction is \emph{fragile} under perturbation of its attention pathways.
By Minsoo Kim, Sungyoung Ji, Kisung Moon, Ilyong Yoon
arXiv:2609.35890v1 Announce Type: new
Abstract: Sparse-autoencoder decomposability and concentrated feature attribution are increasingly treated as evidence that a model's computation is easier to re...
By Adam Elimadi
arXiv:2607. 04223v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) reduces but does not eliminate hallucination, and existing detectors return a single answer-level score that does not indicate which sentence is unsupported, or why.
By Mohamed Aly Bouke
arXiv:2607. 05316v1 Announce Type: cross Abstract: Large language models generate one token at a time, yet their responses show remarkably consistent length structure: step-by-step solutions converge in predictable token counts, retrievals stop after a few sentences, retractions extend responses by measurable amounts.
By Mohamed Amine Merzouk, Dmitri Carpov, Mirko Bronzi, Damiano Fornasiere, Adam Oberman
The paper shows that AI‑text detectors, rather than learning a clear AI‑versus‑human boundary, amplify an inherited predictability axis that already exists in language models. This amplification causes detectors to over‑flag fluent, formal human writing while missing high‑temperature AI outputs, and the bias persists across languages, code, and detector architectures. A training‑free operator can relocate the bias but cannot erase it, underscoring that the unfairness is a structural cost of out‑of‑distribution generalization.
By Alexander Smirnov
arXiv:2609. 31181v1 Announce Type: new Abstract: Black-box model identification works by scoring a model's response to natural-language prompts.
By Nicol\'as Vera Z\'u\~niga
arXiv:2606. 16541v1 Announce Type: new Abstract: Autoformalization, translating natural-language mathematics into formal proof assistants, is bottlenecked not by translation fluency but by \emph{faithfulness}: a formal statement can typecheck and be provable, yet still encode a different theorem than the source intended.
By Noor Islam S. Mohammad, Tamim Sheikh
The study evaluates how extractive prompt compressors affect token costs across ten languages, finding that compressors trained on English data widen the token premium gap for non‑English languages, while a multilingual compressor does not. The gap is tied to the supervision data rather than model architecture, and aggressive compression can reduce non‑English contexts to near‑zero utility. A translate‑then‑compress approach can match or outperform native compression at roughly half the token cost in several languages.
By Mantas Lukauskas
arXiv:2607. 20379v1 Announce Type: new Abstract: Natural-language autoencoders score explanations of hidden activations by reconstruction: an explanation is deemed faithful if the activation can be regenerated from it.
By Hiskias Dingeto