When Context Changes: Understanding Update Failures in LLMs
cs.AI
Submitted: 2026-09-30
Updated: 2026-09-30
Sources
- LLMs Remember First, Forget Last: Dual-Process Interference in Large Language Models
- RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably
- Gemma 2: Improving Open Language Models at a Practical Size
- The Llama 3 Herd of Models
- RULER: What's the Real Context Size of Your Long-Context Language Models?
- Mistral 7B
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model
- A retrieval conditioned rebinding circuit for dynamic entity tracking in large language models
- Supersede: Diagnosing and Training the Memory-Update Gap in LLM Agents
- Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking
- Answer When Needed, Forget When Not: Language Models Pretend to Forget via In-Context Knowledge Unlearning
- Do Language Models Track Entities Across State Changes?
- Steering Language Models With Activation Engineering
- From Recall to Forgetting: Benchmarking Long-Term Memory for Personalized Agents
- Unable to Forget: Proactive Interference Reveals Working Memory Limits in LLMs Beyond Context Length
- Retrieval Head Mechanistically Explains Long-Context Factuality
- DynamicMem: A Long-Horizon Memory Benchmark in Real-World Settings
- Learning to Forget: Sleep-Inspired Memory Consolidation for Resolving Proactive Interference in Large Language Models
- Qwen2.5 Technical Report
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems