Backend
Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts
arXiv stat.ML0 views
arXiv:2609.18366v4 Announce Type: replace-cross Abstract: An optimized agent harness can improve on held-out tasks by relying on formatting conventions that those tasks share with the rest of the benchmark. We introduce Counterfactual Harness Search and Evolution (CHASE) to test whether harness improvements survive changes to the evaluation protocol. A Proposer edits the harness around a fixed foundation model. A Challenger then searches for a protocol change that preserves the tasks but reduces the updated harness's gain over the initial harness. Executable checks test whether the transformation preserves task semantics. If the gain loss is confirmed on held-out tasks, CHASE adds the transformation to an archive that constrains subsequent harness updates. We give conditions under which this finite archive bounds the gain loss on an ideal benchmark with shortcuts removed. We evaluate CHASE on Syn-Ledger, OfficeQA, FinanceBench, and SWE-Bench Pro V2. On Syn-Ledger, removing the known shortcuts yields 38.30% accuracy for CHASE and 18.27% for HarnessEvolve. On the 21 certification tasks from SWE-Bench Pro V2 HARD-51, changing the issue formatting and tool-output field order reduces HarnessEvolve's gain over the initial harness from 8.57 to 3.81 percentage points. The final CHASE harness retains 10.48 points of its 11.43-point gain under the same change.
Source:https://arxiv.org/abs/2609.18366PreviousInverse Design for Conditional Distribution MatchingNextSwiss-Knife: A Framework for Reconfigurable Externalised Multi-Objective Alignment at Decode Time
Related posts

The Transaction Only Covers Your Database. The Provider Key Is What Saves You
0dev.to Database

Go 1.28 will flag the error your rows.Next() loop forgets
0dev.to Database
How zkDatabase Works: From Data Sampling to Verifiable Query Results
0dev.to Database
LogTreeDB: Building a Crash-Safe, Zero-Dependency Document Database from Scratch in C++17
0dev.to Database
CratesBump: A Version-Bump Dataset for the Rust Ecosystem
0arXiv cs.SE
Integrating AI Coding Agents Into Scientific Research
0arXiv cs.SE