Fine-Tuning Actually Worked

Quick follow-up to the college search post. Last time, I gave up on fine-tuning a generative model and built a DistilBERT classifier instead. The classifier works great (100% precision, 100% recall), but I always had this nagging feeling that fine-tuning should have worked if I’d found the right model and the right tooling. Today Liquid AI released LFM2.5-350M, and it changes the picture. ...

March 31, 2026 · 4 min · Franck Verrot

Running a 0.8B Model on an iPhone to Help My Kid Pick a College

College search is a 10-dimensional optimization problem that families solve with spreadsheets, gut feelings, and way too many browser tabs. My kid wants to study biology, play D1 soccer, and ideally not pay more than $40K/year. Oh, and not biomedical, just bio. That’s four constraints already. Add in state preferences, school size, acceptance rates, graduation rates, application systems (Common App? ApplyTexas? UC?), and now you’re juggling variables that don’t fit in your head. ...

March 23, 2026 · 14 min · Franck Verrot

What is the best way to pass NSManagedObjectContext around in iOS applications?

TL;DR Passing an NSManagedObjectContext in each and every NSViewController that needs CoreData can be done by different means, but some are frowned upon. In doubt, inject it. While pair-reviewing some code with my colleague Vincent Tourraine, we realized we wrote this kind of code into each viewDidLoad: method of each NSViewController subclasses we created: managedObjectContext = [[[UIApplication sharedApplication] delegate] managedObjectContext]; Doing so is more or less the same as using a global variable, as calling sharedApplication on UIApplication returns the singleton application instance. I kind of felt this was a code smell so I investigated. ...

January 18, 2012 · 2 min · Franck Verrot