The paper investigates whether targeted edits to a few internal components of Gemma 4 instruction‑tuned models can reduce persistent repetition loops that occur during long factual enumeration prompts. By combining per‑layer ablation with per‑neuron attribution, the authors identify specific neurons whose weight edits dramatically lower loop frequency—one sign‑inverted neuron suffices for Gemma 4 E2B. Across all four Gemma variants, loop occurrences drop from 46/384 to 12/384 on held‑out prompts, while general‑purpose benchmarks show no significant regressions. The study also demonstrates that similar sparse edits can mitigate repetition in other families such as Qwen3.5 and LFM2.5, though the effect varies.
By Aristotelis Lazaridis, Aman Sharma, Dylan Bates, Brian King, Vincent Lu, Jack FitzGerald
arXiv:2511. 05852v4 Announce Type: replace-cross Abstract: Knowledge editing (KE) offers a lightweight alternative to retraining for updating large language models (LLMs).
By Yinjie Cheng, Paul Youssef, Christin Seifert, J\"org Schl\"otterer, Zhixue Zhao
The paper investigates whether knowledge editing truly erases original facts from language models. Using a linear trace probe, the authors find that after editing a fact in GPT‑2‑XL, the original object remains highly decodable from hidden states across three different editing methods, even when the model behaves correctly on edited prompts. This suggests that editing suppresses rather than removes the original association in representational space.
By Priyansh Srivastava, Romit Chatterjee
arXiv:2608.20873v1 Announce Type: new
Abstract: Every way of teaching a deployed language model something new -- full fine-tuning, adapter merging, model editing -- replaces the released checkpoint,...
By Zifeng Liu, Zhiyong Du, Yaxin Lu, Yiming Mao, Zhenhe Wang, Wenqi Shi, Zhengkun Jing
arXiv:2609.25602v1 Announce Type: new
Abstract: In language models, the choice between believing the prompt and believing the weights is made by a handful of identifiable attention heads. Instruction...
By Shubham Santosh Pandere, Gautam Ranka, Ritika Varshney, Navya Deshmukh, Roushni Sareen, Roshan Kumar Singh
arXiv:2605. 12765v3 Announce Type: replace Abstract: Large Language Models memorize vast amounts of training data, raising concerns regarding privacy, copyright infringement, and safety.
By Vin\'icius Conte Turani, Ot\'avio Parraga, Jo\~ao Vitor Boer Abitante, Kristen K. Arguello, Joana Pasquali, Ramiro N. Barros, Flavio du Pin Calmon, Christian Mattjie, Rodrigo C. Barros, Lucas S. Kupssinsk\"u
The paper introduces a fixed‑budget revision protocol that uses deterministic verifiers to expose all remaining violations across exact‑length, lexical, and compositional constraints, thereby isolating model‑side revision behavior. Experiments on 19 open‑ and closed‑source LLMs show wide variability in controller‑level success, with some models achieving up to 99.8% success while others remain below 20%. Controlled studies reveal that post‑training and scale affect model responses to exact feedback, but do not consistently improve exact correction, and that recurrence of earlier outputs is linked to lower recoverability.
By Haitong Jiang, Chunlin Liu, Yile Wang, Yuhong Feng
arXiv:2606. 25449v1 Announce Type: cross Abstract: A language model's memory can be worse than having no memory at all.
By Alex Kwon
arXiv:2606. 20737v2 Announce Type: replace Abstract: We study factual edit propagation in a controlled synthetic knowledge-graph QA setting using a 2x2 grid that crosses loop recurrence with shared-memory access: a dense transformer (Dense), a looped transformer (Loop), a dense backbone with shared memory (Dense+Mem), and a looped backbone with shared memory (loop-memory coupling, LMC).
By Yanan Niu
arXiv:2608. 19893v1 Announce Type: cross Abstract: Where does the novelty a base language model produces with no task come from, and what can an LLM judge of a long stream actually see?
By Roberto I. Ono Filho
The paper investigates how sequential knowledge editing can degrade a language model’s ability to discern reliable evidence from unreliable evidence without affecting overall accuracy. Using a conservatively tuned LoRA on Qwen2.5‑7B‑Instruct, the authors show that after 1,000 edits the model’s arbitration score for untouched facts drops by 36%, leading to higher error rates on its most confident decisions, while MMLU accuracy remains unchanged. The study also finds that in some model‑method combinations, sequential edits can reduce MMLU to chance levels even though edit success and locality remain perfect.
By Atul Anand
The paper investigates the problem of over‑editing by large language models when repairing code, showing that even state‑of‑the‑art models like GPT‑5.5 frequently rewrite more code than necessary. Using a benchmark of 400 BigCodeBench problems with controlled AST corruptions, the authors quantify excess edits and demonstrate that a simple preservation instruction can reduce unnecessary changes and improve pass rates. They further explore training strategies, finding that reinforcement learning yields the best balance between edit fidelity and performance retention, highlighting edit fidelity as a distinct, measurable dimension of code‑repair quality.
By Tongyao Zhu, Wei Hern Lim, Min-Yen Kan