Spaces:
Runtime error
Runtime error
File size: 5,634 Bytes
69ee6e4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 | from __future__ import annotations
import json
from pathlib import Path
from typing import Any
def _provider_table_rows(scored: dict[str, Any]) -> list[tuple[str, dict[str, Any]]]:
return sorted(
scored.get("providers", {}).items(),
key=lambda item: item[1].get("accuracy", 0.0),
reverse=True,
)
def build_markdown_report(run_id: str, scored: dict[str, Any], pairwise: list[dict[str, Any]]) -> str:
lines: list[str] = []
lines.append(f"# Evaluation Report: {run_id}")
lines.append("")
lines.append(f"- Status: `{scored.get('status', 'unknown')}`")
lines.append(f"- Total evaluated rows: `{scored.get('total_rows', 0)}`")
lines.append("")
lines.append("## Leaderboard")
lines.append("")
lines.append("| System | Attempted | Correct | Accuracy | CI95 | Avg Latency (ms) | Errors |")
lines.append("|---|---:|---:|---:|---:|---:|---:|")
for provider, metrics in _provider_table_rows(scored):
ci = metrics.get("accuracy_ci95", {"low": 0.0, "high": 0.0})
lines.append(
"| "
f"{provider} | {metrics.get('attempted', 0)} | {metrics.get('correct', 0)} | "
f"{metrics.get('accuracy', 0.0):.3f} | "
f"[{ci.get('low', 0.0):.3f}, {ci.get('high', 0.0):.3f}] | "
f"{metrics.get('avg_latency_ms', 0.0):.1f} | {metrics.get('errors', 0)} |"
)
lines.append("")
lines.append("## Pairwise Significance (Matched Samples)")
lines.append("")
if not pairwise:
lines.append("No provider pairs available.")
else:
lines.append("| System A | System B | Wins A | Wins B | Ties | p-value |")
lines.append("|---|---|---:|---:|---:|---:|")
for row in pairwise:
lines.append(
f"| {row['provider_a']} | {row['provider_b']} | {row['wins_a']} | "
f"{row['wins_b']} | {row['ties']} | {row['p_value_two_sided']:.4f} |"
)
lines.append("")
lines.append("## Category Breakdown")
lines.append("")
for provider, metrics in _provider_table_rows(scored):
lines.append(f"### {provider}")
lines.append("")
lines.append("| Category | Attempted | Correct | Accuracy |")
lines.append("|---|---:|---:|---:|")
categories = sorted(metrics.get("categories", {}).items(), key=lambda x: x[0])
for cat_name, cat in categories:
lines.append(
f"| {cat_name} | {cat.get('attempted', 0)} | {cat.get('correct', 0)} | "
f"{cat.get('accuracy', 0.0):.3f} |"
)
lines.append("")
return "\n".join(lines).strip() + "\n"
def build_html_report(run_id: str, scored: dict[str, Any], pairwise: list[dict[str, Any]]) -> str:
rows = _provider_table_rows(scored)
provider_rows = []
for provider, metrics in rows:
ci = metrics.get("accuracy_ci95", {"low": 0.0, "high": 0.0})
provider_rows.append(
"<tr>"
f"<td>{provider}</td><td>{metrics.get('attempted', 0)}</td>"
f"<td>{metrics.get('correct', 0)}</td><td>{metrics.get('accuracy', 0.0):.3f}</td>"
f"<td>[{ci.get('low', 0.0):.3f}, {ci.get('high', 0.0):.3f}]</td>"
f"<td>{metrics.get('avg_latency_ms', 0.0):.1f}</td>"
f"<td>{metrics.get('errors', 0)}</td></tr>"
)
pair_rows = []
for row in pairwise:
pair_rows.append(
"<tr>"
f"<td>{row['provider_a']}</td><td>{row['provider_b']}</td>"
f"<td>{row['wins_a']}</td><td>{row['wins_b']}</td><td>{row['ties']}</td>"
f"<td>{row['p_value_two_sided']:.4f}</td></tr>"
)
return (
"<!doctype html><html><head><meta charset='utf-8'>"
f"<title>Evaluation Report {run_id}</title>"
"<style>body{font-family:Arial,sans-serif;margin:24px}table{border-collapse:collapse;width:100%}"
"th,td{border:1px solid #ddd;padding:8px;text-align:left}th{background:#f4f4f4}</style>"
"</head><body>"
f"<h1>Evaluation Report: {run_id}</h1>"
f"<p>Status: <code>{scored.get('status', 'unknown')}</code><br>"
f"Total evaluated rows: <code>{scored.get('total_rows', 0)}</code></p>"
"<h2>Leaderboard</h2><table><thead><tr>"
"<th>System</th><th>Attempted</th><th>Correct</th><th>Accuracy</th>"
"<th>CI95</th><th>Avg Latency (ms)</th><th>Errors</th>"
"</tr></thead><tbody>"
+ "".join(provider_rows)
+ "</tbody></table>"
"<h2>Pairwise Significance</h2><table><thead><tr>"
"<th>System A</th><th>System B</th><th>Wins A</th><th>Wins B</th><th>Ties</th><th>p-value</th>"
"</tr></thead><tbody>"
+ ("".join(pair_rows) if pair_rows else "<tr><td colspan='6'>No pairs available</td></tr>")
+ "</tbody></table>"
"</body></html>"
)
def write_reports(
*,
run_id: str,
scored: dict[str, Any],
pairwise: list[dict[str, Any]],
reports_root: str | Path = "reports",
) -> dict[str, str]:
root = Path(reports_root)
root.mkdir(parents=True, exist_ok=True)
md_path = root / f"{run_id}.md"
html_path = root / f"{run_id}.html"
json_path = root / f"{run_id}.json"
md_path.write_text(build_markdown_report(run_id, scored, pairwise), encoding="utf-8")
html_path.write_text(build_html_report(run_id, scored, pairwise), encoding="utf-8")
json_path.write_text(
json.dumps({"run_id": run_id, "scored": scored, "pairwise": pairwise}, indent=2),
encoding="utf-8",
)
return {"markdown": str(md_path), "html": str(html_path), "json": str(json_path)}
|