NeuroGenesysThe research we build on.
1,514 papers by other researchers that our work cites, from continual learning and verification to language models, agents and hardware. Each entry links to the paper on arXiv and to where we use it. 301 are cited on published pages of this site; the rest in the research notes behind our studies. Our own papers are in the archive.
- 2026-11arXiv 2611.06858CompressionBench: How can your coding agents compress LLMs?
Xinrui Zhong, Zhifeng Li, Zhiming Wang et al. · AI & agents
Recursive self-improvement requires AI systems to improve the models and infrastructure used by subsequent systems. LLM compression offers a concrete efficiency objective: retain a pretrained model's behavior within a smaller model storage budget.
- 2026-10arXiv 2610.22293Towards Continual Learning: Mitigating Forgetting With Subspace Tuning
Arnav Chavan, Kusha Sareen, Rishabh Tiwari et al. · Machine learning
Large language models (LLMs) acquire new skills through fine-tuning, but it can destroy capabilities they already have: catastrophic forgetting remains a central obstacle to continual learning in LLMs.
- 2026-10arXiv 2611.00042Diagnosing Zero-Shot Forgetting in CLIP Fine-Tuning: Attention Drift Characterizes Adaptation, Representation Drift Predicts Transfer Degradation
Ruize Xia · Vision
Fine-tuning CLIP on a downstream task typically reduces its zero-shot accuracy on other tasks, and quantifying this loss ordinarily requires a labeled transfer benchmark.
- 2026-09arXiv 2609.34453SPACE-LoRA: Allocating Activation-Subspace Protection for Continual Learning
Seunghyun Yoo, Kiseok Kim, Hyeontae Joo et al. · Machine learning
This study addresses the catastrophic forgetting problem that occurs when sequentially learning successive tasks using Low-Rank Adaptation (LoRA) from a lifelong learning perspective.
- 2026-09arXiv 2609.29045The Tokens Remember: When Tokenization Bypasses Knowledge Editing and Unlearning
Manit Baser, Aditya Nawal, Dinil Mon Divakaran et al. · Security
Open-weight LLMs give downstream users control over the inference stack, but this flexibility can undermine post-release guarantees that sensitive knowledge has been modified or removed.
- 2026-09arXiv 2609.13680Drift-Constrained Optimization: Only Direction Matters in Fine-Tuning Instruct Models
Fei Yuan, Changjiang Gao, Yilei Tu et al. · AI & agents
Fine-tuning instruct models often improves target performance while inducing behavioral drift from the reference model, which can degrade existing capabilities.
- 2026-09arXiv 2609.09113SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?
Yuqiao Tan, Shizhu He, Jun Zhao et al. · AI & agents
While research on recursive self-improvement (RSI) has predominantly automated model training pipelines, reliable autonomous development demands a missing pillar: post-hoc monitoring and auditing to understand what models learn and ensure safe alignment.
- 2026-08arXiv 2608.25429Distance Is Not Enough: Forget-Retain Alignment Gap Predicts LLM Relearning Robustness
Yi Chen, Hanna Hsieh, Shuhong Liu et al. · Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026), pages 44542-44555 · AI & agents
Machine unlearning aims to make a model forget specific data, yet unlearned LLMs often fail to stay unlearned: brief fine-tuning can revive removed knowledge.
- 2026-08arXiv 2608.26292On Scope Classification and Current Knowledge-Editing Benchmarks: A Negative Result, with INLAY as a Gradient-Free Case Study
Aditya Pratap Singh · Language models & NLP
Every memory-based knowledge editor in the SERAC lineage depends on a scope decision: given a query, does a stored edit apply? We report that current knowledge-editing benchmarks cannot measure this decision at all.
- 2026-08arXiv 2608.15901Layers Matter: Why Continual Learning Regularization Should Be Layer-Adaptive
Brian B. Moser, Ahmed Anwar, Tobias Christian Nauen et al. · Machine learning
Continual learning regularizers like EWC fight forgetting by penalizing changes from previous-task parameters with per-parameter importance, typically diagonal Fisher values.
- 2026-07arXiv 2609.18985Suppressed, Not Erased: A Representational Trace of Edited Facts Survives Even Weight-Free Knowledge Editing
Priyansh Srivastava, Romit Chatterjee · AI & agents
Knowledge-editing benchmarks certify local correctness, whether an edited model produces the new fact on near-edit prompts but not how much of the original fact remains decodable inside the model.
- 2026-07arXiv 2607.11997Are we Merging the Right Models? Impact of Expert Training Duration on Model Merging for LLMs
Nikita Kozodoi, Zainab Afolabi, Jack Butler · Machine learning
Multi-task model merging combines separately trained expert models into a single model that handles all tasks without co-training. Standard practice merges experts at their optimal validation loss.
- 2026-07arXiv 2607.08393Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuning
Lu Dai, Ziyang Rao, Yili Wang et al. · AI & agents
Fine-tuning LLMs to inject new knowledge faces a critical challenge: LLMs can quickly memorize new facts, yet fail to use them for downstream reasoning tasks.
- 2026-06arXiv 2606.24026Can Language Model Agents be Helpful Circuit Explainers in Mechanistic Interpretability?
Ayan Antik Khan, Harsh Kohli, Yuekun Yao et al. · AI & agents
Mechanistic interpretability has made substantial progress in automatically localizing circuits, but explaining what localized components do remains labor-intensive and difficult to standardize.
- 2026-06arXiv 2607.16227LLM Unlearning for Cyber Defense: A Survey on Methods, Challenges, and Emerging Threats
Ruppikha Sree Shankar, Abhishek Bhardwaj, Arnav Doshi et al. · Machine learning
LLMs are increasingly deployed in security-critical systems across healthcare, finance, education, and decision support, yet their inability to forget creates serious cybersecurity, privacy, and safety risks.
- 2026-06arXiv 2606.23276Exposing the Illusion of Erasure in Knowledge Editing for LLMs
Advik Raj Basani, Anshuman Chhabra · Machine learning
Knowledge Editing (KE) has emerged as a frontier for updating specific facts in LLMs without costly retraining, but its reliability and underlying mechanisms remain poorly understood.
- 2026-06arXiv 2606.17168RepSelect: Robust LLM Unlearning via Representation Selectivity
Filip Sondej, Yushi Yang, Adam Mahdi · Language models & NLP
When LLM weights are open or fine-tuning is available through an API, suppressing hazardous knowledge and tendencies is not enough: removal has to be deep enough that an adversary cannot restore it.
- 2026-06arXiv 2606.12883The Hidden Power of Scaling Factor in LoRA Optimization
Zicheng Zhang, Haoran Li, Jiaxing Wang et al. · AI & agents
In Low-Rank Adaptation (LoRA), the scaling factor is often treated as a mere complement to the learning rate, yet its role in optimization remains poorly understood.
- 2026-06arXiv 2606.10554Benchmarking Knowledge Editing using Logical Rules
Tatiana Moteu Ngoli, NDah Jean Kouagou, Hamada M. Zahera et al. · The Semantic Web. ISWC 2025. ISWC 2025. Lecture Notes in Computer Science, vol 16141. Springer, Cham · Language models & NLP
Large Language Models (LLMs) are increasingly deployed in real-world applications that require access to up-to-date knowledge. However, retraining LLMs is computationally expensive.
- 2026-06arXiv 2606.03179HyperPatch: Sequential Knowledge Editing Under n-ary Structural Drift
Yu-Kai Chan, Wen-Sheng Lien, Dong-Ting Yao et al. · Language models & NLP
Large Language Models (LLMs) rely on Knowledge Editing (KE) to maintain temporal validity, yet real-world knowledge is inherently n-ary.
- 2026-06arXiv 2606.03695Don't Forget Your Embeddings: Robust Knowledge Erasure via Precise Editing of Embeddings
Clara Haya Suslik, Or Shafran, Mor Geva · Language models & NLP
As language models are increasingly deployed in real-world applications, the ability to erase specific knowledge from them becomes critical for safety and compliance.
- 2026-06arXiv 2608.12332Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?
Hyowon Wi, Noseong Park · Language models & NLP
In recent years, low-rank adaptation (LoRA) has emerged as a significant paradigm that freezes pre-trained weights and introduces small, learnable adapters instead of fine-tuning the full set of parameters.
- 2026-06arXiv 2606.01610Revisiting Ripple Effects in Knowledge Editing through Pressure-Aware Joint Neighborhood Optimization
Haoben Huang, Shuxin Liu, Ou Wu et al. · AI & agents
Single-edit updates in large language models can trigger ripple effects across local knowledge neighborhoods: desirable propagation to related facts and unintended perturbation of preserved ones.
- 2026-05arXiv 2605.28303From Fact Overwriting to Knowledge Evolution: Causal Editing via On-Policy Self-Distillation
Shuaike Li, Kai Zhang, Xianquan Wang et al. · AI & agents
While Knowledge Editing (KE) enables efficient updates, its dominant Static Fact Overwriting paradigm treats LLMs as discrete databases, forcibly injecting isolated facts.
- 2026-05arXiv 2605.25525SAE-FD: Sparse Autoencoder Feature Distillation for Continual Learning of Large Language Models
Mingxu Zhang, Yuhan Li, Lujundong Li et al. · Machine learning
Continual learning enables large language models to adapt to evolving tasks without retraining from scratch, yet catastrophic forgetting remains a central obstacle.
- 2026-05arXiv 2605.20296Spectral Unforgetting: Post-Hoc Recovery of Damaged Capabilities Without Retraining
Aarash Abro, Muhammad Tahir · Machine learning
Fine-tuning a language model for a target task routinely degrades capabilities the training data never explicitly threatened.
- 2026-05arXiv 2605.18610CATA: Continual Machine Unlearning via Conflict-Averse Task Arithmetic
Shen Lin, Junhao Dong, Rongjie Chen et al. · Vision
Vision-language models (VLMs) have shown remarkable ability in aligning visual and textual representations, enabling a wide range of multimodal applications.
- 2026-05arXiv 2605.16865MixSD: Mixed Contextual Self-Distillation for Knowledge Injection
Jiarui Liu, Lechen Zhang, Yongjin Yang et al. · Language models & NLP
Supervised fine-tuning (SFT) is widely used to inject new knowledge into language models, but it often degrades pretrained capabilities such as reasoning and general-domain performance.
- 2026-05arXiv 2605.11685Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter
Zeguan Xiao, Xuanzhe Xu, Yong Wang et al. · Language models & NLP
Large language model (LLM) unlearning aims to remove specific data influences from pre-trained model without costly retraining, addressing privacy, copyright, and safety concerns.
- 2026-05arXiv 2605.12419ORBIT: Preserving Foundational Language Capabilities in GenRetrieval via Origin-Regulated Merging
Neha Verma, Nikhil Mehta, Shao-Chuan Wang et al. · Language models & NLP
Despite the rapid advancements in large language model (LLM) development, fine-tuning them for specific tasks often results in the catastrophic forgetting of their general, language-based reasoning abilities.
- 2026-05arXiv 2605.12752Low-Rank Adapters Initialization via Gradient Surgery for Continual Learning
Joana Pasquali, Ramiro N. Barros, Arthur S. Bianchessi et al. · Machine learning
LoRA is widely adopted for continual fine-tuning of Large Language Models due to its parameter efficiency, modularity across tasks, and compatibility with replay strategies.
- 2026-05arXiv 2605.07076Self-Consolidating Language Models: Continual Knowledge Incorporation from Context
Zekun Wang, Anant Gupta, Zihan Dong et al. · Language models & NLP
Large language models (LLMs) increasingly receive information as streams of passages, conversations, and long-context workflows. While longer context windows expose more evidence, they do not ensure that useful information is preserved and reused.
- 2026-05arXiv 2605.01555Automated Interpretability and Feature Discovery in Language Models with Agents
Arnau Marin-Llobet, Javier Ferrando · Language models & NLP
We introduce an autonomous multiagent framework for mechanistic interpretability that automates both explaining and finding internal features in large language models.
- 2026-04arXiv 2604.26866MoRFI: Monotonic Sparse Autoencoder Feature Identification
Dimitris Dimakopoulos, Shay B. Cohen, Ioannis Konstas · Language models & NLP
Large language models (LLMs) acquire most of their factual knowledge during the pre-training stage, through next token prediction. Subsequent stages of post-training often introduce new facts outwith the parametric knowledge, giving rise to hallucinations.
- 2026-04arXiv 2604.22407Hidden Failure Modes of Gradient Modification under Adam in Continual Learning, and Adaptive Decoupled Moment Routing as a Repair
Yuelin Hu, Zhenbo Yu, Zhengxue Cheng et al. · Machine learning
, projection, penalty rescaling, replay mixing) while treating Adam as a neutral backend. We show this composition has a hidden failure mode.2).1.8 units.8 units.
- 2026-04arXiv 2604.19089Towards Scalable Lifelong Knowledge Editing with Selective Knowledge Suppression
Dahyun Jung, Jaewook Lee, Heuiseok Lim · AI & agents
Large language models (LLMs) require frequent knowledge updates to reflect changing facts and mitigate hallucinations. To meet this demand, lifelong knowledge editing has emerged as a continual approach to modify specific pieces of knowledge without retraining the entire model.
- 2026-04arXiv 2605.28839One Mask to Rule Them All: On Hidden Facts after Editing and How to Find Them
Ali Holmov, Paul Youssef, Nandi Schoots et al. · Machine learning
Knowledge editing methods such as ROME and MEMIT update factual associations in transformer models by modifying MLP weights. While evaluated mainly by output behavior, their internal mechanism remains underexplored.
- 2026-04arXiv 2604.15574Why Fine-Tuning Encourages Hallucinations and How to Fix It
Guy Kaplan, Zorik Gekhman, Zhen Zhu et al. · Language models & NLP
Large language models are prone to hallucinating factually incorrect statements.~knowledge acquired during pre-training. Since these errors arise as a by-product of knowledge degradation, we explore whether established continual learning tools can mitigate them.
- 2026-03arXiv 2603.20101Pitfalls in Evaluating Interpretability Agents
Tal Haklay, Nikhil Prakash, Sana Pandey et al. · AI & agents
Automated interpretability systems aim to reduce the need for human labor and scale analysis to increasingly large models and diverse tasks.
- 2026-03arXiv 2603.17973TDAD: Test-Driven Agentic Development - Reducing Code Regressions in AI Coding Agents via Graph-Based Impact Analysis
Pepe Alonso, Sergio Yovine, Victor A. Braberman · Society, software & economics
AI coding agents can resolve real-world software issues, yet they frequently introduce regressions -- breaking tests that previously passed. Current benchmarks focus almost exclusively on resolution rate, leaving regression behavior under-studied.
Built from every arXiv paper cited in our papers, posts, topic hubs, reports and research notes, with titles, authors, dates and abstracts from the arXiv API (updated 2026-11-12). Summaries are the opening sentences of each paper's own abstract, not our assessment; a citation means we used or checked the work, not that we endorse it. Fields follow each paper's primary arXiv category. Papers cited only outside arXiv (journals, proceedings) are not listed yet. Entries are generated from our citations, so a mistyped ID in a note would list the wrong paper; we remove those as we find them. Report a wrong entry.
