· Xiaojing Yang · Explainability and Responsible AI
Utility Functions：我们到底在解释哪种模型行为？
Training-data attribution 依赖 utility function。质量、事实性、风格、安全、公平性和术语表现可能指向不同的高影响数据。
Blog
Foundations and research notes on mathematics, statistics, machine learning, NLP, LLMs, model evaluation, and applied AI systems.
- · Xiaojing Yang · Explainability and Responsible AI
什么是 Training Data Attribution？
从 NLP 与 LLM 研究角度解释 training-data attribution：它解释什么、它和 feature attribution 有什么不同，以及为什么它对模型评估和数据中心研究重要。
- · Xiaojing Yang · Statistics
ANOVA for Comparing Multiple Models
ANOVA asks whether group-level variation is larger than within-group noise.
- · Xiaojing Yang · NLP and LLMs
Attention Mechanism
Attention as a learned way to decide what context matters for each token.
- · Xiaojing Yang · Statistics
Bias-Variance Trade-off
Bias and variance explain why both too-simple and too-flexible models can fail.
- · Xiaojing Yang · Machine Learning
Bias–Variance Trade-off in Machine Learning
A practical diagnosis map for underfitting, overfitting, model complexity, and generalization.