Bijan Bowen introduces the LFM3 family and testsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria. a 1.7-billion-parameter text model in a browser interface and the Hermes agent. It performs simple file and interface tasks, but makes mistakes and produces a basic result.
He tests the vision model on photos and a difficult service-manual image. It identifies some content and transcribes much of the table, while missing details. A voice demoA voice interface lets people interact with software through spoken input and audible responses. also makes up information and does not provide true interruption handling.
An omni-model test gives a partially inaccurate video answer, and a final demonstration runs a quantized modelQuantization represents AI model values with fewer numerical levels to reduce memory use or computation cost. on an Arduino Uno Q. Bowen presents the tests as an encouraging but informal look at small models, not a benchmark or guarantee of performance.
Watch on YouTube
