ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
cs.CL, cs.SE
Submitted: 2026-05-22
Updated: 2026-08-27
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning
- Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct
- Refusal in Language Models Is Mediated by a Single Direction
- LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Constitutional AI: Harmlessness from AI Feedback
- Tell me about yourself: LLMs are aware of their learned behaviors
- Looking Inward: Language Models Can Learn About Themselves by Introspection
- From Persona to Personalization: A Survey on Role-Playing Language Agents
- Evaluating Large Language Models Trained on Code
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference
- Examining Identity Drift in Conversations of LLM Agents
- Bias-Augmented Consistency Training Reduces Biased Reasoning in Chain-of-Thought
- Drift No More? Context Equilibria in Multi-Turn LLM Interactions
- Transcoders Find Interpretable LLM Feature Circuits
- SycEval: Evaluating LLM Sycophancy
- Who's asking? User personas and the mechanics of latent misalignment
- A Survey on LLM-as-a-Judge
- LM-Infinite: Zero-Shot Extreme Length Generalization for Large Language Models
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering