-
Automatic Harness Evolution for Hardware Design Verification: Can LLMs Consolidate Gains Across Discovered Harnesses?
We let LLMs evolve the harness around a coding agent for hardware verification. Coverage went way up; correctness improved less, and the gains were split across harnesses instead of consolidating into one.
-
Coding Agents, Harnesses, and Where Things Are Going
Two schools of coding agents, why they're converging, and why the harness — not just the model — is where a lot of the gains now live.
-
MLBox: Automating Machine Learning in Python