Keep K3 suite selection and report-schema scoring in bash, merge K3/vision dataset_args into dpv4 yamls, and pin EvalScope at 735d920ee911 with local patches. Co-authored-by: Cursor <cursoragent@cursor.com>
41 lines
1.3 KiB
Python
41 lines
1.3 KiB
Python
"""Semantic display coverage for the standalone HTML report."""
|
|
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
|
|
from evalscope.report import renderer
|
|
from evalscope.report.report import Report
|
|
|
|
|
|
def test_html_uses_semantic_labels_values_and_raw_name_tooltips(
|
|
tmp_path: Path, monkeypatch: pytest.MonkeyPatch
|
|
) -> None:
|
|
report = Report.from_dict({
|
|
'dataset_name': 'gsm8k',
|
|
'dataset_pretty_name': 'GSM8K',
|
|
'model_name': 'fixture-model',
|
|
'metrics': [{
|
|
'name': 'mean_acc',
|
|
'categories': [{
|
|
'name': ['default'],
|
|
'subsets': [{
|
|
'name': 'test',
|
|
'score': 0.8567,
|
|
'num': 10,
|
|
}],
|
|
}],
|
|
}],
|
|
})
|
|
monkeypatch.setattr(renderer, 'get_report_list', lambda _: [report])
|
|
monkeypatch.setattr(renderer, '_overview_chart_div', lambda _: '')
|
|
monkeypatch.setattr(renderer, '_sunburst_chart_div', lambda _: '')
|
|
monkeypatch.setattr(renderer, '_subset_chart_div', lambda **_: '')
|
|
|
|
output = renderer.gen_html_report_file(str(tmp_path))
|
|
html = Path(output).read_text(encoding='utf-8')
|
|
|
|
assert 'title="accuracy:mean">Accuracy ↑</span>' in html
|
|
assert '<span class="score-pill" data-score="0.8567">85.7%</span>' in html
|
|
assert '<td class="score-cell">85.7%</td>' in html
|