Posts

topics

[Project] 《打開黑盒子：大型語言模型的機制解釋性入門》

Oct 1, 2026

A Traditional Chinese (zh-TW) popular-science series on mechanistic interpretability, with hands-on companion notebooks that let readers explore the methods firsthand.

Interpretability·#project

[Project] ChronoClones: From VR Game to Speculative Fiction

Jan 11, 2025

A VR puzzle game where players solve multi-person challenges alone by collaborating with clones of their past selves through time loops and a speculative fiction inspired by it.

Human-Computer Interaction·#project

All Posts

[Project] 《打開黑盒子：大型語言模型的機制解釋性入門》 — A Traditional Chinese (zh-TW) popular-science series on mechanistic interpretability, with hands-on companion notebooks that let readers explore the methods firsthand.
[Project] ChronoClones: From VR Game to Speculative Fiction — A VR puzzle game where players solve multi-person challenges alone by collaborating with clones of their past selves through time loops and a speculative fiction inspired by it.
[Literature Review] Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior — Review of the paper 'Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior'
[Literature Review] Condition-Dependent Brain-Model Alignment between Whisper and the Human Speech Network under Acoustic Degradation — Review of paired works on how acoustic degradation reweights cortical and ASR processing, measured via layer-wise alignment between Whisper-Tiny and fMRI.
[Thoughts] A Reflection on Taking Chances — A reflection sparked by a coffee chat.
[Literature Review] Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering — Review of the paper 'Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering'
[Literature Review] Parallelized Hierarchical Connectome: A Spatiotemporal Recurrent Framework for Spiking State-Space Models — Review of the paper 'Parallelized Hierarchical Connectome: A Spatiotemporal Recurrent Framework for Spiking State-Space Models'
[Literature Review] EVAAA: A Virtual Environment Platform for Essential Variables in Autonomous and Adaptive Agents — Review of the paper 'EVAAA: A Virtual Environment Platform for Essential Variables in Autonomous and Adaptive Agents'
[Literature Review] The Mind's Transformer: Computational Neuroanatomy of LLM-Brain Alignment — Review of the paper 'The Mind's Transformer: Computational Neuroanatomy of LLM-Brain Alignment'
[Literature Review] Mechanistic Interpretability for Steering Vision-Language-Action Models — Review of the paper 'Mechanistic Interpretability for Steering Vision-Language-Action Models'
[Literature Review] Unveiling Theory of Mind in Large Language Models: A Parallel to Single Neurons in the Human Brain — Review of the paper 'Unveiling Theory of Mind in Large Language Models: A Parallel to Single Neurons in the Human Brain'
[Post-Mortem] OTSea Staking - Lost 26k Due to Logic Flaw — DeFi security incident where OTSeaStaking contract was exploited due to a memory-vs-storage logic flaw, resulting in 26k loss.