44 lines
1.4 KiB
JSON
44 lines
1.4 KiB
JSON
{
|
|
"unit_tests": 18,
|
|
"unit_result": "passed",
|
|
"live_generation_requests": 50,
|
|
"live_generation_errors": 0,
|
|
"source_scenarios_total": 30,
|
|
"pilot_scenarios": 10,
|
|
"modes": [
|
|
"bypass",
|
|
"naive",
|
|
"local",
|
|
"hybrid",
|
|
"mix"
|
|
],
|
|
"pdf_pages": 56,
|
|
"pdf_sha256": "c776084eaf0cdbaa277237fc1a87b870a8ab12dc1a4807e34f20ff0341140bc6",
|
|
"ui": "Query, graph canvas and nonempty answer-source tabs passed Playwright assertions",
|
|
"cli": "Real mix query returned answer and source URLs",
|
|
"limitations": [
|
|
"Pilot is not independent clinical validation",
|
|
"The LLM rubric is not RAGAS; one of 50 latest verdicts was rejected for invalid reference denominator",
|
|
"Official assignment and author declaration remain to be supplied by the author",
|
|
"Native RAGAS uses saved source excerpts, default English prompts, the generator model as judge, and MiniLM semantic inputs truncated to 128 tokens; this is a pilot, not clinical validation"
|
|
],
|
|
"ragas": {
|
|
"version": "0.4.3",
|
|
"answerable_questions": 8,
|
|
"responses": 40,
|
|
"numeric_scores": 144,
|
|
"latest_errors": 0,
|
|
"undefined": 0,
|
|
"skipped": 56,
|
|
"historical_failed_attempts": 1,
|
|
"metrics": [
|
|
"faithfulness",
|
|
"answer_relevancy",
|
|
"context_recall",
|
|
"answer_correctness"
|
|
],
|
|
"pdf_visual_qa": "passed",
|
|
"temporary_embedding_server": "stopped",
|
|
"graph_role_and_metric_interpretation": "clarified; current run scores unchanged"
|
|
}
|
|
} |