engineering.fractional.aiTiny OCRIn our paper reading group last month I presented DeepSeek’s Context Optical Compression paper which claimed amazing performance with a much smaller model (3 billion parameters) than the previous frontier models I had used for OCR (Claude Sonnet and ...Feb 11·3 min read
engineering.fractional.aiOn the new larger context windowsToday the latest LLMs have large context windows up to ~1 million tokens. There are many occasions when this larger context window can be useful: Context engineering: injecting rich system/user context without juggling state Long documents: reading...Nov 4, 2025·2 min read
engineering.fractional.aiLet Computers Talk to ComputersOlder computer systems required humans to adapt their behavior to computers. Instead of describing news articles you were interested in, to search a news index you might have to write Boolean keyword searches: ((“disco” OR “bell bottoms“) AND “Ronald...Apr 2, 2025·3 min read
engineering.fractional.aiUsing Historical Codebase To Build An Eval SuiteZapier has millions of users using thousands of integrations with different platforms. They’re constantly requesting new features they’d like to see, and Zapier has a team dedicated to identifying the most popular and useful, and implementing them to...Nov 11, 2024·3 min read
engineering.fractional.aiFine Tuning QuirkI was working on content moderation for Change.org (see our full case study here). Our goal was to automatically identify violative content to flag for reviewers. The existing system employed several queries, one of which I found had a high percentag...Aug 27, 2024·3 min read
