Tagged “small models”
Research
Can a Small Model Build Slides in a Company's Design Language?
IBM's Palette trains a 20B model to build decks in IBM Design Language. The result is real, but accuracy, editability and portability to other brands are still untested.
Research
When Agent Feedback Makes Small Models Repeat a Failed Action
A paper reports that recording a failed move can make small models more likely to repeat it in games. Here is what that does and doesn't mean for agent retry design.