Reading

The papers my current work rests on, ordered by how much they bear on it. The last column says whether I have checked the venue and year against the paper itself — an unchecked entry came from a secondhand note and should not be cited from here.

Papers: 37
Read: 18
Skimmed: 5
To read: 14
Unchecked: 0

Ordered by how much each bears on the work. A folded corner marks a secondhand entry, not citable yet. Open a cover for the paper.

ICLR2026KnowledgeSmith: Uncovering Knowledge Updating in LLMs with Model Editing and UnlearningLuo et al.To read ICLR 20262025Bilinear representation mitigates reversal curse and enables consistent model editingKim et al.To read Findings of EMNLP 20252025STEAM: A Semantic-Level Knowledge Editing Framework for Large Language ModelsJeong et al.To read arXiv preprint2025The Curious Case of Factual (Mis)Alignment between LLMs' Short- and Long-Form AnswersObaid et al.To read arXiv 2512.041442025RippleBench: Capturing Ripple Effects Using Existing Knowledge RepositoriesRinberg et al.To read arXiv 2605.124262026Geometric Factual Recall in TransformersRavfogel et al.Skimmed arXiv 2601.046002026On the Limitations of Rank-One Model Editing in Answering Multi-hop QuestionsChen et al.Skimmed Preprint2025ChainEdit: Propagating Ripple Effects in LLM Knowledge Editing through Logical Rule-Guided ChainsDong et al.✓ Read ICLR2025NNsight and NDIF: Democratizing Access to Open-Weight Foundation Model InternalsFiotto-Kaufman et al.✓ Read arXiv preprint2025Diagnosing Model Editing via Knowledge Spectrum(preprint & authors✓ Read TACL2024Evaluating the Ripple Effects of Knowledge Editing in Language ModelsCohen et al.✓ Read ICLR2024Function Vectors in Large Language ModelsTodd et al.✓ Read ICLR2023Mass-Editing Memory in a TransformerMeng et al.✓ Read EMNLP2023MQuAKE: Assessing Knowledge Editing in Language Models via Multi-Hop QuestionsZhong et al.✓ Read NeurIPS2023Does Localization Inform Editing? Surprising Differences in Causality-Based Localization vs. Knowledge EditingHase et al.✓ Read NeurIPS2022Locating and Editing Factual Associations in GPTMeng et al.✓ Read EMNLP2021Surface Form Competition: Why the Highest Probability Answer Isn't Always RightHoltzman et al.✓ Read TACL2021Measuring and Improving Consistency in Pretrained Language ModelsElazar et al.✓ Read arXiv preprint2026Revisiting Ripple Effects in Knowledge Editing through Pressure-Aware Joint Neighborhood OptimizationHuang et al.To read ICLR2026Tracing and Reversing Edits in LLMsYoussef et al.To read Findings of ACL2026Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language ModelsZhang et al.To read arXiv 2606.028602026Forgetting is Not Erasure: Recovering Latent Knowledge via Transport KeysChaudhuryTo read ICML (oral)2026AI Engram: In Search of Memory Traces in Artificial IntelligenceKwon et al.To read arXiv 2607.248052026Forgetting Is Not a Fix: Path Dependence in Sequential Engram EditingSchesslTo read ICML 20252024Representation Shattering in Transformers: A Synthetic Study with Knowledge EditingNishi et al.To read ACL2024RAVEL: Evaluating Interpretability Methods on Disentangling Language Model RepresentationsHuang et al.To read COLM2024The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False DatasetsMarks & Tegmark✓ Read EMNLP2023Editing Large Language Models: Problems, Methods, and OpportunitiesYao et al.✓ Read EMNLP2022Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary SpaceGeva et al.✓ Read ACL2022Knowledge Neurons in Pretrained TransformersDai et al.✓ Read EMNLP2021Transformer Feed-Forward Layers Are Key-Value MemoriesGeva et al.✓ Read arXiv 2605.020832026EditPropBench: Measuring Factual Edit Propagation in Scientific ManuscriptsKruthofTo read arXiv 2309.12499 (cs.SE)2023CodePlan: Repository-level Coding using LLMs and PlanningBairi et al.Skimmed ICML2022Memory-Based Model Editing at ScaleMitchell et al.Skimmed arXiv preprint2020Modifying Memories in Transformer ModelsZhu et al.Skimmed NeurIPS2023Aging with GRACE: Lifelong Model Editing with Discrete Key-Value AdaptorsHartvigsen et al.✓ Read ICLR2022Fast Model Editing at ScaleMitchell et al.✓ Read

Kept in data/papers.yml. In any piece, <Cite id="cohen2024ripple" /> resolves to the entry.