← All IntelClip / Developer ToolsWhy existing benchmarks fail: clustering, brittle verifiers, leakage
From DeepSWE: A Contamination-Resistant Coding Benchmark — James Shi, Datacurve · ≈2:40
“contamination is also rampant because again all of these tasks are mined from public PRs.”
“The verifiers are also very very brittle because we're anchoring them to a specific implementation often derived from the PR that was merged in.”
What’s in it
- Critiques SWE-bench Pro's flaws for evaluating coding agents
- Explains why Deep Suite was built to fix benchmark contamination
- Reveals how models like Claude game benchmarks via git log
Clip transcript
initiatives that we have towards answering this question. So why did we create Deep Suite? Well, it was very clear that the existing benchmarks are not hitting the mark. Um with benches like Sweetbench Pro, uh top models are clustering at the top. It's very hard to differentiate between uh which one is good because they all have overlapping confidence intervals. uh contamination is also rampant because again all of these tasks are mined from public PRs. So all the solution tests even the discussion around the PRs those are all available out in the wild for these agents to access. The verifiers are also very very brittle because we're anchoring them to a specific implementation often derived from the PR that was merged in. And oftentimes you also have tests that check for private helpers and functions uh created by the task author which is uh very opinionated right and is not something that models should have to adhere to. Um and finally uh leakage. So one thing about Sweetbench Pro is uh for very uh insightful models such as Claude, they're able to directly run git log and then go through the commit hashes and cherrypick the ones out that contain the golden patches which again very very serious issue.
Comments
Sign in to comment.
Loading comments…