File size: 11,852 Bytes
3487f22 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 | """
Visualize eval_metrics.json results. Charts are saved to the charts/ subdirectory.
Produces:
charts/overall_scores.png β bar chart of avg score per scorer
charts/category_heatmap.png β heatmap of avg score by category Γ scorer
charts/latency_distribution.png β per-case latency score distribution
charts/failures_by_category.png β count of sub-threshold cases per category
Usage:
python agent-evaluation/visualize_metrics.py
python agent-evaluation/visualize_metrics.py --metrics eval_metrics.json --threshold 1.0
"""
import argparse
import json
from pathlib import Path
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import numpy as np
DEFAULT_METRICS_PATH = Path(__file__).resolve().parent / "eval_metrics.json"
CHARTS_DIR = Path(__file__).resolve().parent / "charts"
DEFAULT_THRESHOLD = 1.0
PALETTE = {
"pass": "#4CAF50",
"warn": "#FF9800",
"fail": "#F44336",
"bar": "#5C6BC0",
"bg": "#FAFAFA",
}
_SCORER_LABELS = {
"BenchmarkSimilarityValidity": "Benchmark Similarity Validity",
"ClassificationValidity": "Classification Validity",
"ClauseStructureValidity": "Clause Structure Validity",
"ExpectedClauseType": "Expected Clause Type",
"Factuality": "Factuality",
"Latency": "Latency",
"NoError": "No Error",
"OutputStructureValidity": "Output Structure Validity",
"RiskFactorsPresence": "Risk Factors Presence",
"RiskScoreValidity": "Risk Score Validity",
}
_CATEGORY_LABELS = {
"agency_agreement": "Agency Agreement",
"commercial_lease": "Commercial Lease",
"consulting_agreement": "Consulting Agreement",
"data_processing_agreement": "Data Processing Agreement",
"distribution_agreement": "Distribution Agreement",
"distributor_agreement": "Distributor Agreement",
"employment_contract": "Employment Contract",
"enterprise_software": "Enterprise Software",
"franchise_agreement": "Franchise Agreement",
"joint_venture": "Joint Venture",
"master_service_agreement": "Master Service Agreement",
"nda": "NDA",
"nda_services_agreement": "NDA Services Agreement",
"partnership_agreement": "Partnership Agreement",
"research_collaboration": "Research Collaboration",
"reseller_agreement": "Reseller Agreement",
"saas_agreement": "SaaS Agreement",
"services_agreement": "Services Agreement",
"software_license": "Software License",
"strategic_alliance": "Strategic Alliance",
"supply_agreement": "Supply Agreement",
"technology_licensing": "Technology Licensing",
"technology_transfer": "Technology Transfer",
}
def _fmt(name: str, mapping: dict) -> str:
return mapping.get(name, name.replace("_", " ").title())
def load_metrics(path: str) -> dict:
p = Path(path)
if not p.exists():
raise FileNotFoundError(f"Metrics file not found: {path}")
with open(p) as f:
return json.load(f)
def _save(fig: plt.Figure, name: str) -> None:
CHARTS_DIR.mkdir(parents=True, exist_ok=True)
out = CHARTS_DIR / name
fig.savefig(out, bbox_inches="tight", dpi=150)
plt.close(fig)
print(f" saved β {out}")
# ---------------------------------------------------------------------------
# Chart 1: Overall scores bar chart
# ---------------------------------------------------------------------------
def plot_overall_scores(metrics: dict) -> None:
overall = metrics.get("overall_scores", {})
scorers = sorted(overall.keys())
averages = [overall[s]["average"] for s in scorers]
counts = [overall[s]["count"] for s in scorers]
colors = [
PALETTE["pass"] if v == 1.0
else PALETTE["warn"] if v >= 0.75
else PALETTE["fail"]
for v in averages
]
fig, ax = plt.subplots(figsize=(10, 5))
fig.patch.set_facecolor(PALETTE["bg"])
ax.set_facecolor(PALETTE["bg"])
scorer_labels = [_fmt(s, _SCORER_LABELS) for s in scorers]
bars = ax.barh(scorer_labels, averages, color=colors, edgecolor="white", height=0.6)
for bar, avg, n in zip(bars, averages, counts):
ax.text(
min(avg + 0.005, 1.01), bar.get_y() + bar.get_height() / 2,
f"{avg:.3f} (n={n})",
va="center", ha="left", fontsize=9, color="#333333",
)
ax.set_xlim(0, 1.12)
ax.set_xlabel("Average Score", fontsize=11)
ax.set_title("Overall Scorer Averages", fontsize=13, fontweight="bold", pad=12)
ax.axvline(1.0, color="#BDBDBD", linewidth=0.8, linestyle="--")
ax.tick_params(axis="y", labelsize=10)
ax.spines[["top", "right", "bottom"]].set_visible(False)
total = metrics.get("total_cases", 0)
ax.text(
0.99, -0.08, f"n = {total} total cases",
transform=ax.transAxes, ha="right", fontsize=9, color="#757575",
)
_save(fig, "overall_scores.png")
# ---------------------------------------------------------------------------
# Chart 2: Category Γ Scorer heatmap
# ---------------------------------------------------------------------------
def plot_category_heatmap(metrics: dict) -> None:
per_category = metrics.get("per_category", {})
categories = sorted(per_category.keys())
all_scorers: set[str] = set()
for cat_data in per_category.values():
all_scorers.update(cat_data.keys())
scorers = sorted(all_scorers)
matrix = np.full((len(categories), len(scorers)), np.nan)
for r, cat in enumerate(categories):
for c, scorer in enumerate(scorers):
entry = per_category[cat].get(scorer)
if entry is not None:
matrix[r, c] = entry["average"]
fig, ax = plt.subplots(figsize=(max(10, len(scorers) * 1.3), max(6, len(categories) * 0.55)))
fig.patch.set_facecolor(PALETTE["bg"])
ax.set_facecolor(PALETTE["bg"])
cmap = mcolors.LinearSegmentedColormap.from_list(
"eval", [PALETTE["fail"], PALETTE["warn"], PALETTE["pass"]]
)
cmap.set_bad(color="#E0E0E0")
im = ax.imshow(matrix, cmap=cmap, vmin=0.0, vmax=1.0, aspect="auto")
ax.set_xticks(range(len(scorers)))
ax.set_xticklabels([_fmt(s, _SCORER_LABELS) for s in scorers], rotation=35, ha="right", fontsize=9)
ax.set_yticks(range(len(categories)))
ax.set_yticklabels([_fmt(c, _CATEGORY_LABELS) for c in categories], fontsize=9)
for r in range(len(categories)):
for c in range(len(scorers)):
val = matrix[r, c]
if not np.isnan(val):
text_color = "white" if val < 0.6 else "#222222"
ax.text(c, r, f"{val:.2f}", ha="center", va="center",
fontsize=7.5, color=text_color)
plt.colorbar(im, ax=ax, fraction=0.02, pad=0.02, label="Average Score")
ax.set_title("Score Heatmap: Category Γ Scorer", fontsize=13, fontweight="bold", pad=12)
ax.tick_params(length=0)
_save(fig, "category_heatmap.png")
# ---------------------------------------------------------------------------
# Chart 3: Per-case Latency score distribution
# ---------------------------------------------------------------------------
def plot_latency_distribution(metrics: dict) -> None:
latency_scores = [
case["scores"].get("Latency")
for case in metrics.get("per_case", [])
if case.get("scores", {}).get("Latency") is not None
]
if not latency_scores:
print(" no latency scores found β skipping latency chart")
return
buckets = {1.0: 0, 0.75: 0, 0.5: 0, 0.25: 0, 0.0: 0}
for s in latency_scores:
buckets[s] = buckets.get(s, 0) + 1
labels = {
1.0: "< 30s\n(1.0)",
0.75: "30β60s\n(0.75)",
0.5: "60β120s\n(0.5)",
0.25: "120β180s\n(0.25)",
0.0: "> 180s\n(0.0)",
}
ordered = [1.0, 0.75, 0.5, 0.25, 0.0]
counts = [buckets.get(v, 0) for v in ordered]
tick_labels = [labels[v] for v in ordered]
colors = [PALETTE["pass"], PALETTE["warn"], PALETTE["warn"], PALETTE["fail"], PALETTE["fail"]]
fig, ax = plt.subplots(figsize=(8, 4))
fig.patch.set_facecolor(PALETTE["bg"])
ax.set_facecolor(PALETTE["bg"])
bars = ax.bar(tick_labels, counts, color=colors, edgecolor="white", width=0.55)
for bar, count in zip(bars, counts):
if count:
ax.text(
bar.get_x() + bar.get_width() / 2, bar.get_height() + 0.15,
str(count), ha="center", va="bottom", fontsize=10,
)
ax.set_ylabel("Number of Cases", fontsize=11)
ax.set_title("Latency Score Distribution (per case)", fontsize=13, fontweight="bold", pad=12)
ax.spines[["top", "right", "left"]].set_visible(False)
ax.tick_params(axis="x", length=0)
ax.set_yticks([])
_save(fig, "latency_distribution.png")
# ---------------------------------------------------------------------------
# Chart 4: Failures per category (cases below threshold)
# ---------------------------------------------------------------------------
def plot_failures_by_category(metrics: dict, threshold: float) -> None:
failure_counts: dict[str, int] = {}
for case in metrics.get("per_case", []):
cat = case.get("category", "unknown")
has_failure = any(
s < threshold for s in case.get("scores", {}).values()
) or bool(case.get("error"))
if has_failure:
failure_counts[cat] = failure_counts.get(cat, 0) + 1
if not failure_counts:
print(f" no failures at threshold {threshold} β skipping failures chart")
return
categories = sorted(failure_counts, key=lambda c: -failure_counts[c])
counts = [failure_counts[c] for c in categories]
category_labels = [_fmt(c, _CATEGORY_LABELS) for c in categories]
fig, ax = plt.subplots(figsize=(9, max(4, len(categories) * 0.5)))
fig.patch.set_facecolor(PALETTE["bg"])
ax.set_facecolor(PALETTE["bg"])
ax.barh(category_labels, counts, color=PALETTE["fail"], edgecolor="white", height=0.6)
for i, (cat, count) in enumerate(zip(category_labels, counts)):
ax.text(count + 0.05, i, str(count), va="center", fontsize=10)
ax.set_xlabel("Number of Failed Cases", fontsize=11)
ax.set_title(
f"Failures by Category (threshold < {threshold})",
fontsize=13, fontweight="bold", pad=12,
)
ax.spines[["top", "right", "bottom"]].set_visible(False)
ax.set_xlim(0, max(counts) + 1.5)
ax.tick_params(axis="y", labelsize=9)
_save(fig, "failures_by_category.png")
# ---------------------------------------------------------------------------
# CLI
# ---------------------------------------------------------------------------
def main() -> None:
parser = argparse.ArgumentParser(
description="Visualize eval_metrics.json. Charts saved to charts/",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog=__doc__,
)
parser.add_argument(
"--metrics", type=str, default=str(DEFAULT_METRICS_PATH),
help=f"Path to eval_metrics.json (default: {DEFAULT_METRICS_PATH})",
)
parser.add_argument(
"--threshold", type=float, default=DEFAULT_THRESHOLD,
help=f"Failure threshold for the failures-by-category chart (default: {DEFAULT_THRESHOLD})",
)
args = parser.parse_args()
metrics = load_metrics(args.metrics)
print(f"\nGenerating charts from {args.metrics} β {CHARTS_DIR}/\n")
plot_overall_scores(metrics)
plot_category_heatmap(metrics)
plot_latency_distribution(metrics)
plot_failures_by_category(metrics, args.threshold)
print("\nDone.\n")
if __name__ == "__main__":
main()
|