ContextEcho benchmark shows persona drift occurs across 23 frontier models in long agentic-coding sessions, is not reliably reset by compaction, and can be restored by single-shot anchors with mode-dependent effects.
On the emergence of position bias in transformers
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
LLMs show strong user bias in role-tagged contexts that is amplified by preference alignment and can be reduced or controlled through targeted fine-tuning and DPO.
A training-free attention-guided debiasing framework mitigates position bias in MLLM multi-image retrieval by exploiting the observed mismatch between biased logits and aligned attention maps, yielding over 40% accuracy gains on MS-COCO benchmarks.
S4 models exhibit stable time-continuity unlike sensitive S6 models, with task continuity predicting performance and enabling temporal subsampling for better efficiency.
LLMs show a quality-dependent position bias, favoring the first option for high-quality choices and later options for low-quality ones, and higher-temperature sampling can reveal the underlying preference.
Empirical Jacobian analysis reveals that token influence in trained language models decays as a power law with distance (exponent ~0.8), a learned property not present in random models.
A 1650-session factorial study found no measurable impact from config file size, instruction position, architecture, or conflicts on coding agent adherence, though compliance declined within sessions.
Irminsul recovers up to 83% of prompt tokens above exact-prefix matching and delivers 63% prefill energy savings per cache hit on MLA-MoE models by content-hashing CDC chunks and applying closed-form kr correction.
SDSR places human metadata at file primacy and combines it with prompt routing rules to reach 100% primary category accuracy on a 119-category benchmark, far above the 65% no-guidance baseline.
citing papers explorer
-
ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
ContextEcho benchmark shows persona drift occurs across 23 frontier models in long agentic-coding sessions, is not reliably reset by compaction, and can be restored by single-shot anchors with mode-dependent effects.
-
User-Assistant Bias in LLMs
LLMs show strong user bias in role-tagged contexts that is amplified by preference alignment and can be reduced or controlled through targeted fine-tuning and DPO.
-
Logit-Attention Divergence: Mitigating Position Bias in Multi-Image Retrieval via Attention-Guided Calibration
A training-free attention-guided debiasing framework mitigates position bias in MLLM multi-image retrieval by exploiting the observed mismatch between biased logits and aligned attention maps, yielding over 40% accuracy gains on MS-COCO benchmarks.
-
Continuity Laws for Sequential Models
S4 models exhibit stable time-continuity unlike sensitive S6 models, with task continuity predicting performance and enabling temporal subsampling for better efficiency.
-
Fragile Preferences: A Deep Dive Into Order Effects in Large Language Models
LLMs show a quality-dependent position bias, favoring the first option for high-quality choices and later options for low-quality ones, and higher-temperature sampling can reveal the underlying preference.
-
How Token Influence Decays with Distance: A Green-Function View of Trained Language Models
Empirical Jacobian analysis reveals that token influence in trained language models decays as a power law with distance (exponent ~0.8), a learned property not present in random models.
-
Instruction Adherence in Coding Agent Configuration Files: A Factorial Study of Four File-Structure Variables
A 1650-session factorial study found no measurable impact from config file size, instruction position, architecture, or conflicts on coding agent adherence, though compliance declined within sessions.
-
Irminsul: MLA-Native Position-Independent Caching for Agentic LLM Serving
Irminsul recovers up to 83% of prompt tokens above exact-prefix matching and delivers 63% prefill energy savings per cache hit on MLA-MoE models by content-hashing CDC chunks and applying closed-form kr correction.
-
Self-Describing Structured Data with Dual-Layer Guidance: A Lightweight Alternative to RAG for Precision Retrieval in Large-Scale LLM Knowledge Navigation
SDSR places human metadata at file primacy and combines it with prompt routing rules to reach 100% primary category accuracy on a 119-category benchmark, far above the 65% no-guidance baseline.