Search results

Consider pinning public-skill context in benchmark metadata

Status: Open.

#744 In OpenHands/benchmarks;

· neubig opened on Jun 9, 2026

SWE-Bench Lite Docker workspace fails due to image tag validation mismatch

Status: Open.

#733 In OpenHands/benchmarks;

· sm2909 opened on Jun 2, 2026

Port SWE-Bench Multimodal benchmark to Harbor

Status: Open.

#726 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Port SWEfficiency benchmark to Harbor

Status: Open.

#725 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Port HybridGym benchmarks to Harbor

Status: Open.

#724 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Port ProgramBench benchmark to Harbor

Status: Open.

#723 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Port OpenAgentSafety benchmark to Harbor

Status: Open.

#722 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Port Commit0 benchmark to Harbor

Status: Open.

#721 In OpenHands/benchmarks;

· neubig opened on May 28, 2026

Assess impact of swtbench non-test patch stripping on historical runs

Status: Open.

#718 In OpenHands/benchmarks;

· juanmichelini opened on May 27, 2026

swtbench: qwen3-coder-next score is artificially low — agent writes source-code fix alongside the test (78% of patches)

build-swt-benchBuild SWT-Bench images based on SDK version on this PR.Build SWT-Bench images based on SDK version on this PR.

investigation

Status: Open.

#708 In OpenHands/benchmarks;

· juanmichelini opened on May 12, 2026

Difference in models results

Status: Open.

#707 In OpenHands/benchmarks;

· juanmichelini opened on May 12, 2026

Track strict offline isolation for ProgramBench inference

Status: Open.

#706 In OpenHands/benchmarks;

· neubig opened on May 10, 2026