An SLM executor under tight constraints can drive a Cognitive Agent through formatted outputs and logit-based action selection, yielding partial rather than full explainability¶

Claim · All claims
Evidence · 1 study · design r2 · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q1 i? The authors report that an SLM with sufficient performance capacity can execute the minimal, formatted outputs required to drive the Cognitive Agent within tightly constrained action spaces, making certain state transitions more inspectable, though this should not be interpreted as full explainability. → Hyeongdon Moon 2026

Evidence¶

Hyeongdon Moon 2026¶

Hyeongdon Moon, Carolyn Rosé, and John Stamper. (2026). Cognitive Agent Compilation for Explicit Problem Solver Modeling. https://arxiv.org/abs/2605.07040 doi:10.48550/arxiv.2605.07040

q1 · i? · design · r2

Authors' preliminary observation from the SLM proof-of-concept run (Section 5.2), not a validated empirical comparison. They report that enforcing "predefined output formats and leveraging logit-based signals" makes certain state transitions more inspectable, while cautioning this is not full explainability. No effect sizes are printed.

"The results suggest that an SLM with sufficient performance capacity can execute the minimal, formatted outputs required to drive the Cognitive Agent within tightly constrained action spaces. By enforcing the generation of predefined output formats and leveraging logit-based signals, the system allows certain state transitions to become more inspectable."

Discussion¶

Related Claims¶

• Model size defines a core trade-off: too-large SLMs answer without the cognitive structure while too-small SLMs break instruction-following coherence — related
• A 4-bit quantized 27B local Teacher LLM generated knowledge in the required format, suggesting accessible-resource implementations may be feasible
• Underlying all AI risks in education is that generative AI behavior is difficult to explain, inspect, or predict due to its statistical mechanisms — related