grant-radar / evals /eval_set.json
Riley Coleman
feat: add working QA endpoint and fix evaluation harness
94fcd9b
Raw
History Blame Contribute Delete
1.49 kB
{
"eval_cases": [
{
"id": "eval_001",
"category": "search",
"difficulty": "easy",
"query": "innovation funding",
"competition_id": "competition-2265",
"expected_answer": "grants",
"expected_reasoning_contains": [
"found"
],
"min_citations": 1
},
{
"id": "eval_002",
"category": "search",
"difficulty": "easy",
"query": "innovate uk",
"competition_id": "competition-2275",
"expected_answer": "innovation",
"expected_reasoning_contains": [
"found"
],
"min_citations": 1
},
{
"id": "eval_003",
"category": "search",
"difficulty": "medium",
"query": "quantum",
"competition_id": "competition-2288",
"expected_answer": "quantum",
"expected_reasoning_contains": [
"found"
],
"min_citations": 1
},
{
"id": "eval_004",
"category": "search",
"difficulty": "medium",
"query": "horizon",
"competition_id": "competition-1389",
"expected_answer": "horizon",
"expected_reasoning_contains": [
"found"
],
"min_citations": 1
},
{
"id": "eval_005",
"category": "search",
"difficulty": "hard",
"query": "uk grants",
"competition_id": "competition-2265",
"expected_answer": "grant",
"expected_reasoning_contains": [
"found"
],
"min_citations": 1
}
]
}