Selected work

Representative Papers

Gotta Catch them all: the modes of Sycophancy

Jul 22, 2026

This study addresses the phenomenon of "flattery" in large language models—where outputs prioritize alignment with user beliefs over factual accuracy—previously misconstrued as a monolithic behavior. By analyzing 948 social pressure scenarios, the work reveals that flattery actually comprises three distinct modes, differing fundamentally in representational structure and computational mechanisms. Integrating textual classification, internal representation analysis, attention circuit tracing, and cross-layer linear separability tests, the research demonstrates that although the three flattery types produce highly similar outputs (achieving only 57.8% classification accuracy), their internal representations become fully linearly separable from layer 14 onward, with divergent activation dynamics and input preference patterns. These findings challenge the prevailing one-dimensional view of flattery, uncovering its intrinsic heterogeneity and mechanistic separability within transformer architectures.

0 citationsRead paper

Real-Time Drivers' Drowsiness Detection and Analysis through Deep Learning

Nov 15, 2025

To address road safety risks posed by driver fatigue, this paper proposes a non-intrusive, low-cost, real-time detection method leveraging a deep convolutional neural network (DCNN) and OpenCV. The system captures facial video streams via an in-vehicle camera, employs a pre-trained model combined with facial landmark detection to accurately quantify fatigue indicators—specifically PERCLOS (Percentage of Eyelid Closure) and yawning—and triggers multi-level alerts accordingly. Evaluated on the NTHU-DDD and Yawn-Eye-Dataset benchmarks, the method achieves 99.6% and 97.0% classification accuracy for fatigue states, respectively, outperforming existing lightweight approaches. Our key contributions include: (1) a fully end-to-end, deployable real-time framework; (2) a balanced design achieving high accuracy with low computational overhead; and (3) successful integration into an intelligent in-vehicle platform, with robustness and sub-second response latency empirically validated under real-world driving conditions.

0 citationsRead paper

Recent publications

Latest Papers

Gotta Catch them all: the modes of Sycophancy

Jul 22, 2026

This study addresses the phenomenon of "flattery" in large language models—where outputs prioritize alignment with user beliefs over factual accuracy—previously misconstrued as a monolithic behavior. By analyzing 948 social pressure scenarios, the work reveals that flattery actually comprises three distinct modes, differing fundamentally in representational structure and computational mechanisms. Integrating textual classification, internal representation analysis, attention circuit tracing, and cross-layer linear separability tests, the research demonstrates that although the three flattery types produce highly similar outputs (achieving only 57.8% classification accuracy), their internal representations become fully linearly separable from layer 14 onward, with divergent activation dynamics and input preference patterns. These findings challenge the prevailing one-dimensional view of flattery, uncovering its intrinsic heterogeneity and mechanistic separability within transformer architectures.

0 citationsRead paper

Real-Time Drivers' Drowsiness Detection and Analysis through Deep Learning

Nov 15, 2025

To address road safety risks posed by driver fatigue, this paper proposes a non-intrusive, low-cost, real-time detection method leveraging a deep convolutional neural network (DCNN) and OpenCV. The system captures facial video streams via an in-vehicle camera, employs a pre-trained model combined with facial landmark detection to accurately quantify fatigue indicators—specifically PERCLOS (Percentage of Eyelid Closure) and yawning—and triggers multi-level alerts accordingly. Evaluated on the NTHU-DDD and Yawn-Eye-Dataset benchmarks, the method achieves 99.6% and 97.0% classification accuracy for fatigue states, respectively, outperforming existing lightweight approaches. Our key contributions include: (1) a fully end-to-end, deployable real-time framework; (2) a balanced design achieving high accuracy with low computational overhead; and (3) successful integration into an intelligent in-vehicle platform, with robustness and sub-second response latency empirically validated under real-world driving conditions.

0 citationsRead paper