CantHarm Supplementary Summary

Main benchmark summary

Representative result matrix

Task	Metric	Model	Value
Form coarse classification	Macro-F1	XLM-R	0.5353
Form fine classification	Macro-F1	Hierarchical TF-IDF + Linear SVM	0.4594
Form severity regression	Spearman	char TF-IDF + Ridge	0.6037
Fusion coarse classification	Macro-F1	Tuned late fusion	0.5415
Sense coarse classification	Macro-F1	TF-IDF + LogReg	0.5033
Sense fine classification	Macro-F1	TF-IDF + LogReg	0.4057

Reading frame

How these anchors are meant to be used

Anchor family	Why it stays foregrounded
Text and sense tasks	They summarize the main lexical and ambiguity-sensitive benchmark results.
Severity prediction	It remains part of the core manuscript narrative and is interpretable within the benchmark rubric.
Fusion coarse classification	It provides a compact speech-grounded extension in the public summary.

How to read this page

Representative summary

This page groups the main benchmark results reported in the manuscript.
The public metadata bundle is aligned to the same release snapshot.
Task definitions for the full benchmark suite remain documented in Protocol.

Bundle note. This supplementary page is aligned to the current public release snapshot. For the exact workbook, manifest, statistics, benchmark highlights, and speaker-packaged audio files, use Downloads.