II.
Benchmark JSON
Structured · livebenchmark:mbpp
MBPP json
Inspect the normalized record payload exactly as the atlas UI reads it.
{
"id": "benchmark:mbpp",
"_kind": "Benchmark",
"_file": "benchmarks/benchmarks/mbpp.yaml",
"_cluster": "benchmarks",
"attributes": {
"displayName": "MBPP",
"homepageUrl": "https://github.com/google-research/google-research/tree/master/mbpp",
"kind": "code-generation",
"targetsKind": "ModelVersion",
"description": "Mostly Basic Python Problems — ~1,000 short Python programming problems with tests.\n"
},
"outgoingEdges": [],
"incomingEdges": [
{
"from": "eval-result:mbpp.qwen-2-5-coder-32b.001",
"to": "benchmark:mbpp",
"kind": "scored_against",
"attributes": {}
},
{
"from": "eval-run:mbpp.qwen-2-5-coder-32b.2024-11",
"to": "benchmark:mbpp",
"kind": "for_benchmark"
},
{
"from": "scope-boundary:mbpp.scope",
"to": "benchmark:mbpp",
"kind": "bounds_subject"
}
]
}