Reward writes goals into the optimizer, but it does not prove the model learned the right goals. This beginner-oriented rewrite explains why post-reward verification is needed.
XDU -> WHU
Data Science
LLM Agents
Evaluation/Benchmark
iOS Dev
Game Client Dev
-
AEnvironment: Why Agent Development Needs an Interaction Environment Layer
SWE-bench, SWE-agent, and Tau-bench show that agent development cannot focus only on models and frameworks. Tasks, tools, users, state, rules, and validators shape whether agents work.
-
Reinforcement Learning in LLM Alignment: From Reward Signals to Advantage Estimation
Following the signal chain from reward to baseline, advantage, and normalization, this post explains why RL algorithms for LLM alignment keep rewriting reward signals.
-
Why Better Simulators Often Combine Learning and Rules: From PDEs and Ray Tracing to DLSS
Better simulators often come from a clear division of labor: encode conservation, geometry, causality, boundaries, and rendering structure, then let learning cover expensive or fuzzy parts.
-
From MCP to Agent Skills: Why Agents Need a New Context Engineering Protocol
Agent Skills are popular because they package capabilities with low friction. They are a useful layer of context engineering, but not a new universal protocol or final agent form.
-
From RL Agents to LLM Agents: Paradigm Shift and Uncertainty Modeling After The Second Half
The shift from RL agents to LLM agents is not a simple model replacement; learning signals move from numeric spaces into language spaces, reshaping priors, evaluation, and memory.
-
From Agent Cognitive Architecture to Agent Frameworks
Starting from CoALA’s cognitive architecture, this post discusses Workflow, Agent, Supervisor, Agent Team, and MAS boundaries, plus the engineering value of frameworks such as LangGraph.
-
Making Agents Practical: Structured Output and Constrained Decoding for LLMs
Why is structured output hard for agents? This post explains what prompts, JSON mode, constrained decoding, and inference engines each solve.
-
Why Data Keeps Fooling You: The Counterintuitive Inspection Paradox
Why does your wait for a bus often feel longer than the published average interval? This post explains the inspection paradox, a common statistical trap behind such intuitions.
-
Why Language Models Hallucinate
A reading of OpenAI’s Why Language Models Hallucinate, arguing that hallucination is tied to training paradigms and binary evaluation pressure, not only data noise or model flaws.