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.
- 2024-06arXiv 2406.14026Demystifying Language Model Forgetting with Low-rank Example Associations
Xisen Jin, Xiang Ren · Machine learning
Large language models (LLMs) suffer from forgetting of upstream knowledge when fine-tuned. Despite efforts on mitigating forgetting, few have investigated how forgotten upstream examples are dependent on newly learned tasks.
- 2024-06arXiv 2406.01436Editing the Mind of Giants: An In-Depth Exploration of Pitfalls of Knowledge Editing in Large Language Models
Cheng-Hsun Hsueh, Paul Kuo-Ming Huang, Tzu-Han Lin et al. · Language models & NLP
Knowledge editing is a rising technique for efficiently updating factual knowledge in large language models (LLMs) with minimal alteration of parameters.
- 2024-05arXiv 2405.09673LoRA Learns Less and Forgets Less
Dan Biderman, Jacob Portes, Jose Javier Gonzalez Ortiz et al. · Machine learning
Low-Rank Adaptation (LoRA) is a widely-used parameter-efficient finetuning method for large language models. LoRA saves memory by training only low rank perturbations to selected weight matrices.
- 2024-05arXiv 2405.07518SambaNova SN40L: Scaling the AI Memory Wall with Dataflow and Composition of Experts
Raghu Prabhakar, Ram Sivaramakrishnan, Darshan Gandhi et al. · Hardware, photonics & physics
Monolithic large language models (LLMs) like GPT-4 have paved the way for modern generative AI applications. Training, serving, and maintaining monolithic LLMs at scale, however, remains prohibitively expensive and challenging.
- 2024-04arXiv 2404.10198ClashEval: Quantifying the tug-of-war between an LLM's internal prior and external evidence
Kevin Wu, Eric Wu, James Zou · Language models & NLP
Retrieval augmented generation (RAG) is frequently used to mitigate hallucinations and provide up-to-date knowledge for large language models (LLMs).
Cited inStable wrong beliefs evade every self-check: a literature synthesis on model self-knowledge and confident errorsTo keep what a model knows, control what it is allowed to moveWhat a model lets into long-term memory matters more than how it learnsStudy: Fortifying knowledge in continual learningStudy: Hardware guardian core for continual learningStudy: Memory world models and source trustStudy: Model self knowledge and confident errors - 2024-03arXiv 2403.08763Simple and Scalable Strategies to Continually Pre-train Large Language Models
Adam Ibrahim, Benjamin Thérien, Kshitij Gupta et al. · Machine learning
Large language models (LLMs) are routinely pre-trained on billions of tokens, only to start the process over again once new data becomes available. A much more efficient solution is to continually pre-train these models, saving significant compute compared to re-training.
Cited inTo keep what a model knows, control what it is allowed to moveStudy: Elastic weight consolidation deep diveStudy: Fortifying knowledge in continual learningStudy: Hardware guardian core for continual learningStudy: Hierarchical and T shaped MoEStudy: Orchestrated specialists and bridge synthesis - 2024-03arXiv 2403.01244Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal
Jianheng Huang, Leyang Cui, Ante Wang et al. · Language models & NLP
Large language models (LLMs) suffer from catastrophic forgetting during continual learning. Conventional rehearsal-based methods rely on previous training data to retain the model's ability, which may not be feasible in real-world applications.
- 2024-02arXiv 2402.07867PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models
Wei Zou, Runpeng Geng, Binghui Wang et al. · Security
Large language models (LLMs) have achieved remarkable success due to their exceptional generative capabilities. Despite their success, they also have inherent limitations such as a lack of up-to-date knowledge and hallucination.
- 2024-01arXiv 2401.08565Tuning Language Models by Proxy
Alisa Liu, Xiaochuang Han, Yizhong Wang et al. · Language models & NLP
Despite the general capabilities of large pretrained language models, they consistently benefit from further adaptation to better achieve desired behaviors. However, tuning these models has become increasingly resource-intensive, or impossible when model weights are private.
- 2024-01arXiv 2401.07453Model Editing at Scale leads to Gradual and Catastrophic Forgetting
Akshat Gupta, Anurag Rao, Gopala Anumanchipalli · Language models & NLP
Editing knowledge in large language models is an attractive capability to have which allows us to correct incorrectly learnt facts during pre-training, as well as update the model with an ever-growing list of new facts.
- 2024-01arXiv 2401.05605Scaling Laws for Forgetting When Fine-Tuning Large Language Models
Damjan Kalajdzievski · Language models & NLP
We study and quantify the problem of forgetting when fine-tuning pre-trained large language models (LLMs) on a downstream task. We find that parameter-efficient fine-tuning (PEFT) strategies, such as Low-Rank Adapters (LoRA), still suffer from catastrophic forgetting.
- 2024-01arXiv 2401.05566Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
Evan Hubinger, Carson Denison, Jesse Mu et al. · Security
Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity.
- 2024-01arXiv 2401.02415LLaMA Pro: Progressive LLaMA with Block Expansion
Chengyue Wu, Yukang Gan, Yixiao Ge et al. · Language models & NLP
, from LLaMA to CodeLLaMA. To this end, we propose a new post-pretraining method for LLMs with an expansion of Transformer blocks. We tune the expanded blocks using only new corpus, efficiently and effectively improving the model's knowledge without catastrophic forgetting.
- 2023-11arXiv 2311.08106Carpe Diem: On the Evaluation of World Knowledge in Lifelong Language Models
Yujin Kim, Jaehong Yoon, Seonghyeon Ye et al. · Language models & NLP
The dynamic nature of knowledge in an ever-changing world presents challenges for language models trained on static data; the model in the real world often requires not only acquiring new knowledge but also overwriting outdated information into updated ones.
- 2023-10arXiv 2310.14152Orthogonal Subspace Learning for Language Model Continual Learning
Xiao Wang, Tianze Chen, Qiming Ge et al. · Language models & NLP
Benefiting from massive corpora and advanced hardware, large language models (LLMs) exhibit remarkable capabilities in language understanding and generation. However, their performance degrades in scenarios where multiple tasks are encountered sequentially, also known as catastrophic forgetting.
Cited inEWC protects what its Fisher sees: a literature synthesis on elastic weight consolidation, scale and coherenceStudy: Continual learning order and companion dataStudy: Elastic weight consolidation deep diveStudy: Fortifying knowledge in continual learningStudy: Genesys program prior artStudy: Hardware guardian core for continual learningStudy: Merging and splitting MoE expertsStudy: Orchestrated specialists and bridge synthesisStudy: Seven continual learning methods tested - 2023-10arXiv 2310.12962An Emulator for Fine-Tuning Large Language Models using Small Language Models
Eric Mitchell, Rafael Rafailov, Archit Sharma et al. · Language models & NLP
Widely used language models (LMs) are typically built by scaling up a two-stage training pipeline: a pre-training stage that uses a very large, diverse dataset of text and a fine-tuning (sometimes, 'alignment') stage that uses targeted examples or other specifications of desired behaviors.
- 2023-10arXiv 2310.06762TRACE: A Comprehensive Benchmark for Continual Learning in Large Language Models
Xiao Wang, Yuansen Zhang, Tianze Chen et al. · Language models & NLP
Aligned large language models (LLMs) demonstrate exceptional capabilities in task-solving, following instructions, and ensuring safety. However, the continual learning aspect of these aligned LLMs has been largely overlooked.
- 2023-08arXiv 2308.03296Studying Large Language Model Generalization with Influence Functions
Roger Grosse, Juhan Bae, Cem Anil et al. · Machine learning
When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of evidence is: which training examples most contribute to a given behavior?
Cited inCoherence is selected, anchored, then pruned: a literature synthesis on logical coherence in LLM world modelsEWC protects what its Fisher sees: a literature synthesis on elastic weight consolidation, scale and coherenceStudy: Decision signatures for conflict detectionStudy: Elastic weight consolidation deep diveStudy: Hardware guardian core for continual learningStudy: Logical coherence of LLM world models - 2023-07arXiv 2307.12976Evaluating the Ripple Effects of Knowledge Editing in Language Models
Roi Cohen, Eden Biran, Ori Yoran et al. · Language models & NLP
Modern language models capture a large body of factual knowledge. However, some facts can be incorrectly induced or become obsolete over time, resulting in factually incorrect generations.
Cited inA gradient kernel flags contradictions but does not predict forgetting, and steering rehearsal by it backfiresCoherence is selected, anchored, then pruned: a literature synthesis on logical coherence in LLM world modelsEWC protects what its Fisher sees: a literature synthesis on elastic weight consolidation, scale and coherenceIterated self-closure settles fast, but only evidence that overrides self-belief makes the settled state trueTo stop a model forgetting, rehearse what it knew. New context won't do it.Study: Consequence consistent updatesStudy: Continual learning order and companion dataStudy: Decision signatures for conflict detectionStudy: Elastic weight consolidation deep diveStudy: Graded succession by earned authorityStudy: Hardware guardian core for continual learningStudy: Logical coherence of LLM world modelsStudy: Orchestrated specialists and bridge synthesis - 2023-06arXiv 2306.12142Synaptic metaplasticity with multi-level memristive devices
Simone D'Agostino, Filippo Moro, Tifenn Hirtzlin et al. · Neuroscience & neural computation
Deep learning has made remarkable progress in various tasks, surpassing human performance in some cases. However, one drawback of neural networks is catastrophic forgetting, where a network trained on one task forgets the solution when learning a new one.
- 2023-05arXiv 2305.12281Lifelong Language Pretraining with Distribution-Specialized Experts
Wuyang Chen, Yanqi Zhou, Nan Du et al. · Language models & NLP
Pretraining on a large-scale corpus has become a standard method to build general language models (LMs). Adapting a model to new data distributions targeting different downstream tasks poses significant challenges.
- 2022-12arXiv 2212.02872A 64-core mixed-signal in-memory compute chip based on phase-change memory for deep neural network inference
Manuel Le Gallo, Riduan Khaddam-Aljameh, Milos Stanisavljevic et al. · Nature Electronics 6, 680-693 (2023) · Hardware, photonics & physics
The need to repeatedly shuttle around synaptic weight values from memory to processing units has been a key source of energy inefficiency associated with hardware implementation of artificial neural networks.
- 2022-06arXiv 2206.04016SYNERgy between SYNaptic consolidation and Experience Replay for general continual learning
Fahad Sarfraz, Elahe Arani, Bahram Zonooz · Neuroscience & neural computation
Continual learning (CL) in the brain is facilitated by a complex set of mechanisms.
- 2022-04arXiv 2204.14211TemporalWiki: A Lifelong Benchmark for Training and Evaluating Ever-Evolving Language Models
Joel Jang, Seonghyeon Ye, Changho Lee et al. · Language models & NLP
Language Models (LMs) become outdated as the world changes; they often fail to perform tasks requiring recent factual information which was absent or different during training, a phenomenon called temporal misalignment.
- 2022-03arXiv 2203.02155Training language models to follow instructions with human feedback
Long Ouyang, Jeff Wu, Xu Jiang et al. · Language models & NLP
Making language models bigger does not inherently make them better at following a user's intent. For example, large language models can generate outputs that are untruthful, toxic, or simply not helpful to the user.
- 2022-01arXiv 2201.12604Learning Fast, Learning Slow: A General Continual Learning Method based on Complementary Learning System
Elahe Arani, Fahad Sarfraz, Bahram Zonooz · Machine learning
Humans excel at continually learning from an ever-changing environment whereas it remains a challenge for deep neural networks which exhibit catastrophic forgetting.
- 2021-09arXiv 2109.01903Robust fine-tuning of zero-shot models
Mitchell Wortsman, Gabriel Ilharco, Jong Wook Kim et al. · Vision
, without fine-tuning on a specific dataset). Although existing fine-tuning methods substantially improve accuracy on a given target distribution, they often reduce robustness to distribution shifts.
- 2021-05arXiv 2105.03048Regression Bugs Are In Your Model! Measuring, Reducing and Analyzing Regressions In NLP Model Updates
Yuqing Xie, Yi-an Lai, Yuanjun Xiong et al. · Language models & NLP
Behavior of deep neural networks can be inconsistent between different versions. Regressions during model update are a common cause of concern that often over-weigh the benefits in accuracy or efficiency gain.
- 2021-03arXiv 2103.09762Gradient Projection Memory for Continual Learning
Gobinda Saha, Isha Garg, Kaushik Roy · International Conference on Learning Representations (ICLR), 2021 · Machine learning
The ability to learn continually without forgetting the past tasks is a desired attribute for artificial learning systems. Existing approaches to enable such learning in artificial neural networks usually rely on network growth, importance based weight update or replay of old data from the memory.
- 2021-02arXiv 2102.11174Linear Transformers Are Secretly Fast Weight Programmers
Imanol Schlag, Kazuki Irie, Jürgen Schmidhuber · Machine learning
We show the formal equivalence of linearised self-attention mechanisms and fast weight controllers from the early '90s, where a ``slow" neural net learns by gradient descent to program the ``fast weights" of another net through sequences of elementary programming instructions which are additive outer products of…
- 2020-01arXiv 2001.08361Scaling Laws for Neural Language Models
Jared Kaplan, Sam McCandlish, Tom Henighan et al. · Machine learning
We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power-law with model size, dataset size, and the amount of compute used for training, with some trends spanning more than seven orders of magnitude.
- 2019-10arXiv 1910.07104Orthogonal Gradient Descent for Continual Learning
Mehrdad Farajtabar, Navid Azizan, Alex Mott et al. · Machine learning
Neural networks are achieving state of the art and sometimes super-human performance on learning tasks across a variety of domains.
- 2017-06arXiv 1706.08840Gradient Episodic Memory for Continual Learning
David Lopez-Paz, Marc'Aurelio Ranzato · Machine learning
One major obstacle towards AI is the poor ability of models to solve new problems quicker, and without forgetting previously acquired knowledge.
- 2016-10arXiv 1610.06258Using Fast Weights to Attend to the Recent Past
Jimmy Ba, Geoffrey Hinton, Volodymyr Mnih et al. · Machine learning
Until recently, research on artificial neural networks was largely restricted to systems with only two types of variable: Neural activities that represent the current or recent input and weights that learn to capture regularities among inputs, outputs and payoffs.
- 2016-06arXiv 1606.09282Learning without Forgetting
Zhizhong Li, Derek Hoiem · Vision
When building a unified vision system or gradually adding new capabilities to a system, the usual assumption is that training data for all tasks is always available.
- 2016-06arXiv 1606.04671Progressive Neural Networks
Andrei A. Rusu, Neil C. Rabinowitz, Guillaume Desjardins et al. · Machine learning
Learning to solve complex sequences of tasks--while both leveraging transfer and avoiding catastrophic forgetting--remains a key obstacle to achieving human-level intelligence.
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.
