Hyacehila's Blog

Hypothesis testing is powerful but assumption-bound; statistical significance does not make a conclusion true, and anyone without statistical literacy can accidentally commit statistical malpractice.

Why are boosted trees still such strong baselines for structured tabular data? This post compares XGBoost, LightGBM, and CatBoost by technical focus and best-fit use cases.

A technical overview of the path from language-only models to vision-language models, covering CLIP, VLMs, diffusion guidance, and native multimodal tokenization.

Starting from high-dimensional, small-sample settings and the curse of dimensionality, this post frames dimensionality reduction alongside EDA, regularization, and sparsity before surveying PCA/MDS, t-SNE, UMAP, and autoencoders.

Are calculations precise while charts are rough? Francis Anscombe’s quartet challenges that belief and shows why exploratory data analysis matters in statistical inference.

This post revisits C. R. Rao’s statistical classic, reviewing the history of statistics and the problems that emerged as the discipline developed.

LLMs make access to knowledge, explanation, translation, and expression cheaper than ever, but equal access to answers is not the same as equal understanding, judgment, or learning paths.

A personal reflection on 2025 and the two decades before it: what I did, the choices that brought me here, and what I still want to do.

Walks through PocketFlow’s BaseNode, Node, Flow, Batch, and Async implementations to explain how a minimal LLM agent framework organizes workflows with graphs, shared state, and node protocols.