LUKE articles | lilting channel

TechFeb 28, 202615 min

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.

NLP OCR Machine Learning Python BERT LUKE Ollama LLM WSL2 NDLOCR-Lite Experiment

#LUKE

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