Generalization Is Stability, Not Accuracy: Multi-Axis Evaluation of LLMs
SAGO scores LLM generalization as stability across input variants, not single-format accuracy.
1 stories, newest first · one card per story
SAGO scores LLM generalization as stability across input variants, not single-format accuracy.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views