[Submitted on 12 Oct 2026 (v1), last revised 30 Oct 2026 (this version, v2)]
Abstract:The pass-all-test binary reward has become a standard recipe for training software engineering agents, but its sparsity eliminates difference of partially correct patches. The test pass-rate is a natural alternative. However, we find that it is not a reliable indicator of patch quality. A functionality being tested can be covered by either one challenging test or one hundred trivial tests, where the majority might be already satisfied pass-to-pass tests. We introduce CodeLift, a reinforcement learning method that dynamically calibrates pass-rate rewards, turning the test suite into an adaptive curriculum. At each iteration, CodeLift groups tests that kill the same patches and patches that pass the same tests, then gives more weight to test groups that fewer patch groups pass. This reduces the impact of majority simple tests and focuses more on what is still challenging. Within each rollout, CodeLift applies the same scoring rule to repository snapshots, assigning credit to useful actions even when later ones introduce regressions. Experiments on SWE benchmarks show that CodeLift consistently improves pass@1/2/4/8 over binary and pass-rate baselines. Analyses show that CodeLift provides dense, stable learning signals and unlocks the inherent curriculum in the test suite.
Submission history
From: Junjielong Xu [view email]
[v1] Mon, 12 Oct 2026 18:00:58 UTC (1,041 KB)
[v2] Fri, 30 Oct 2026 09:23:23 UTC (1,041 KB)
