#Python

Automated OCR Error Detection and Correction with Encoder Models + Local LLM

Experiment log: from LUKE/BERT fill-mask fine-tuning, to perplexity-based error detection, to Qwen2.5 7B correction judgment with human escalation on mismatch. A complete pipeline running on a single RTX 4060 Laptop with 8GB VRAM.

Building a Local OCR Hot Folder for Confidential Documents with ScanSnap + NDLOCR-Lite

A folder-watching script that automatically OCRs images scanned by ScanSnap and runs LLM-based correction. Includes air-gap security design.

Three Months with NDLOCR: The Full Journey and What Others Built

From Docker hell to Lite + LLM correction. A retrospective on my own experimentation, plus an introduction to someone else's browser-based NDLOCR-Lite implementation.

OCR Correction on Showa-Era Documents with NDLOCR-Lite and Local LLMs

Set up the CLI version of NDLOCR-Lite on Apple Silicon Mac, then tested OCR result correction with Qwen 3.5 and Swallow. Includes experiments with direct image reading and the anchoring effect.

Running the National Diet Library's OCR Engine NDLOCR-Lite on Windows

Tried the lightweight OCR tool NDLOCR-Lite released by the National Diet Library — installed it on Windows 11 and tested both the CLI and GUI versions.

With gr.HTML in Gradio 6, You Can Build a Full Web App from a Single Python File

This article explains how Gradio 6's new gr.HTML component works. You can write HTML, CSS, and JavaScript directly inside Python and build interactive web apps without a separate build step.

ACE-Step: an open-source music generation model that runs locally

A look at ACE-Step, the 'Stable Diffusion of music,' covering its architecture, features, installation, and expected performance on Apple Silicon before trying it on an M1 Max.

Agent Lightning: Microsoft's reinforcement learning framework for AI agents

Microsoft released an open-source framework that can optimize almost any AI agent with reinforcement learning, with little to no code changes. It supports arbitrary frameworks such as LangChain, AutoGen, and Claude Agent SDK.

MarkItDown - Microsoft's Python tool for turning documents into Markdown

A Microsoft tool that converts PDFs, Word, Excel, PowerPoint, and more into Markdown. It also fits into LLM pipelines and supports an MCP server.

PageIndex - tree RAG with LLM reasoning only, no vector search

I looked into PageIndex, a RAG system that builds hierarchical document trees using only LLM reasoning, without chunking or vector databases. I also consider how it fits with layout detection and OCR pipelines.

JAXA Earth API for Python: a simple way to work with satellite data

A look at JAXA's Python package for Earth observation satellite data, covering installation, basic usage, and Claude Desktop integration.

Setting Up LLM/RAG for Work, So I Tried LoRA Training on the Side (Part 1)

Introducing a Mac mini M4 Pro to build an in-house RAG system. A plan for setting up a LoRA training environment during downtime while waiting for specs to be finalized.