Characterizing Model-Native Skills
Quick summary
arXiv:2604.17614v2 Announce Type: replace Abstract: Skills are a natural unit for describing what a language model can do and how its behavior can be changed. However, existing characterizations rely on human-written taxonomies, textual descriptions, or manual profiling pipelines--all external hypotheses about what matters that need not align with the model's internal representations. We argue that when the goal is to intervene on model behavior, skill characterization should be *model-native*: grounded in the model's own representations rather than imposed through external ontologies. We inst
Key takeaways
- arXiv:2604.17614v2 Announce Type: replace Abstract: Skills are a natural unit for describing what a language model can do and how its behavior can be changed.
- However, existing characterizations rely on human-written taxonomies, textual descriptions, or manual profiling pipelines--all external hypotheses about what matters that need not align with the model's internal representations.
- We argue that when the goal is to intervene on model behavior, skill characterization should be *model-native*: grounded in the model's own representations rather than imposed through external ontologies.
Why it matters
“Characterizing Model-Native Skills” should be evaluated beyond branding and benchmark scores. Its practical importance will emerge in task accuracy, latency, unit cost, safety and integration with real workflows.
Member comments