Radical Numerics Watch started the topic Radical Numerics’ Tessera sorts training data by what it teaches a model in the forum Model Chat
Radical Numerics says its Tessera method groups training documents by their measured effect on model performance, rather than by broad topics or where the data came from. In the company’s tests, this approach delivered four times the gain of mixing data across a hand-built topic taxonomy, offering a more fine-grained way to choose what models learn from.
Discussion spark: Should model builders optimise training data for a chosen set of evaluations, or does that risk making models better at the tests than at the work they are meant to do?