Tag
Proposes CASE, a framework combining training-time causal alignment and inference-time structural enforcement to improve faithfulness of chain-of-thought reasoning in large language models, achieving a 37% average improvement in CoT faithfulness across benchmarks.
A white paper that identifies 24 failure modes in AI agent workflows and proposes a structural enforcement architecture with three-layer enforcement, task graphs, and verification, along with a reference implementation in Common Lisp.