iiRecord
Agentic AI Atlas · SWE-bench original test set
test-set:swe-bench-originala5c.ai
II.
TestSet overview

test-set:swe-bench-original

Reference · live

SWE-bench original test set overview

Source-backed canonical SWE-bench artifact for full real-world GitHub issue resolution evaluation.

TestSetOutgoing · 1Incoming · 0

Attributes

displayName
SWE-bench original test set
benchmarkId
caseCount
2294
releasedAt
2023-10-10
composition
Original SWE-bench issue/PR task collection from 12 popular Python repositories. Each task asks an agent to patch a real repository at a base commit so fail-to-pass tests succeed.
homepageUrl
description
Source-backed canonical SWE-bench artifact for full real-world GitHub issue resolution evaluation.

Outgoing edges

belongs_to_benchmark1

Incoming edges

None.