evalarc / comparison /comparison.json
glayguo's picture
Publish verified EvalArc 67dc74806725
1dd51b8 verified
Raw History Blame Contribute Delete
2.48 kB
{
"schema_version": "evalarc.comparison.v1",
"created_at": "2026-09-14T01:58:21.259930+00:00",
"task": {
"id": "support-routing",
"version": "0.1.0",
"domain": "business-tools",
"split": "public-development"
},
"grader_sha256": "2dced2a842369299e6e924db626fedc120071d5731cd9b79202dba7395b90255",
"cases_sha256": "b7cd8663c69da9e1715d5c8e1d707514863dcc333fefc57acc3315053ebc85df",
"runtime": {
"backend": "docker",
"image": "python:3.12-slim",
"image_id": "sha256:78387bc3881b8273120a12ebe6c1ab22b018ccc2c9adf565ae1ac9b536e184ea",
"command": [
"{python}",
"-I",
"-B",
"main.py"
],
"response_timeout_seconds": 10.0,
"session_output_limit_bytes": 1048576,
"python": "3.12.3",
"platform": "Linux-6.17.0-1020-aws-x86_64-with-glibc2.39"
},
"seeds": [
17
],
"baseline": {
"candidate_sha256": "a34429419d7c794420d43f7873ce5b1474f73821329f414609d5e4d18fc174c3",
"score": 0.9,
"resolved": false,
"created_at": "2026-09-14T01:58:19.087946+00:00"
},
"current": {
"candidate_sha256": "b89a99819c03e8290096ca268f637c5517d465b3ec5f0b5fe2b21c35509ff202",
"score": 0.9375,
"resolved": false,
"created_at": "2026-09-14T01:58:21.160793+00:00"
},
"score_delta": 0.0375,
"regressions": [
{
"seed": 17,
"case_id": "retry-after-commit",
"check": "notes"
}
],
"improvements": [
{
"seed": 17,
"case_id": "retry-before-commit",
"check": "closure"
},
{
"seed": 17,
"case_id": "route-open-ticket",
"check": "closure"
}
],
"case_transitions": [
{
"seed": 17,
"case_id": "retry-after-commit",
"before": "passed",
"after": "failed",
"regressed_checks": [
"notes"
],
"improved_checks": []
},
{
"seed": 17,
"case_id": "retry-before-commit",
"before": "failed",
"after": "passed",
"regressed_checks": [],
"improved_checks": [
"closure"
]
},
{
"seed": 17,
"case_id": "route-open-ticket",
"before": "failed",
"after": "passed",
"regressed_checks": [],
"improved_checks": [
"closure"
]
}
],
"current_failed_cases": 1,
"has_regressions": true,
"interpretation": "Matched observed checks only. No regression is not full task resolution. Scores and fingerprints do not authenticate a report's producer."
}