Skip to content
← Back to explorer

Taiwan Safety Benchmark and Breeze Guard: Toward Trustworthy AI for Taiwanese Mandarin

Po-Chun Hsu, Meng-Hsi Chen, Tsu Ling Chao, Chia Tien Han, Da-shan Shiu · Mar 7, 2026 · Citations: 0

Automatic Metrics
Open arXiv Find Implementation RSS feed Shortlist (0)

Data freshness

Extraction: Fresh

Check recency before relying on this page for active eval decisions. Use stale pages as context and verify against current hub results.

Metadata refreshed

Mar 7, 2026, 5:13 PM

Recent

Extraction refreshed

Mar 13, 2026, 9:28 PM

Fresh

Extraction source

Persisted extraction

Confidence 0.45

Abstract

Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin. This limitation results in systematic blind spots when interpreting region-specific risks such as localized financial scams, culturally embedded hate speech, and misinformation patterns. To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin. TS-Bench contains 400 human-curated prompts spanning critical domains including financial fraud, medical misinformation, social discrimination, and political manipulation. In parallel, we present Breeze Guard, an 8B safety model derived from Breeze 2, our previously released general-purpose Taiwanese Mandarin LLM with strong cultural grounding from its original pre-training corpus. Breeze Guard is obtained through supervised fine-tuning on a large-scale, human-verified synthesized dataset targeting Taiwan-specific harms. Our central hypothesis is that effective safety detection requires the cultural grounding already present in the base model; safety fine-tuning alone is insufficient to introduce new socio linguistic knowledge from scratch. Empirically, Breeze Guard significantly outperforms the leading 8B general-purpose safety model, Granite Guardian 3.3, on TS-Bench (+0.17 overall F1), with particularly large gains in high-context categories such as scam (+0.66 F1) and financial malpractice (+0.43 F1). While the model shows slightly lower performance on English-centric benchmarks (ToxicChat, AegisSafetyTest), this tradeoff is expected for a regionally specialized safety model optimized for Taiwanese Mandarin. Together, Breeze Guard and TS-Bench establish a new foundation for trustworthy AI deployment in Taiwan.

Low-signal caution for protocol decisions

Use this page for context, then validate protocol choices against stronger HFEPX references before implementation decisions.

  • Extraction flags indicate low-signal or possible false-positive protocol mapping.
  • Extraction confidence is 0.45 (below strong-reference threshold).
Human Eval Hub LLM-as-Judge Hub Pairwise Preference Hub

HFEPX Relevance Assessment

This paper is adjacent to HFEPX scope and is best used for background context, not as a primary protocol reference.

Best use

Background context only

Use if you need

A benchmark-and-metrics comparison anchor.

Main weakness

Extraction flags indicate low-signal or possible false-positive protocol mapping.

Trust level

Low

Eval-Fit Score

5/100 • Low

Treat as adjacent context, not a core eval-method reference.

Human Feedback Signal

Not explicit in abstract metadata

Evaluation Signal

Detected

HFEPX Fit

Adjacent candidate

Extraction confidence: Low

If you are doing eval pipeline work, start here:

Human Eval Hub LLM-as-Judge Hub Pairwise Preference Hub Tool-Use Eval Hub

Field Provenance & Confidence

Each key protocol field shows extraction state, confidence band, and data source so you can decide whether to trust it directly or validate from full text.

Human Feedback Types

missing

None explicit

Confidence: Low Source: Persisted extraction missing

No explicit feedback protocol extracted.

Evidence snippet: Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin.

Evaluation Modes

partial

Automatic Metrics

Confidence: Low Source: Persisted extraction evidenced

Includes extracted eval setup.

Evidence snippet: Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin.

Quality Controls

missing

Not reported

Confidence: Low Source: Persisted extraction missing

No explicit QC controls found.

Evidence snippet: Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin.

Benchmarks / Datasets

partial

Ts Bench

Confidence: Low Source: Persisted extraction evidenced

Useful for quick benchmark comparison.

Evidence snippet: To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin.

Reported Metrics

partial

F1

Confidence: Low Source: Persisted extraction evidenced

Useful for evaluation criteria comparison.

Evidence snippet: Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin.

Rater Population

missing

Unknown

Confidence: Low Source: Persisted extraction missing

Rater source not explicitly reported.

Evidence snippet: Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin.

Human Data Lens

  • Uses human feedback: No
  • Feedback types: None
  • Rater population: Unknown
  • Unit of annotation: Unknown
  • Expertise required: Medicine
  • Extraction source: Persisted extraction

Evaluation Lens

  • Evaluation modes: Automatic Metrics
  • Agentic eval: None
  • Quality controls: Not reported
  • Confidence: 0.45
  • Flags: low_signal, possible_false_positive

Protocol And Measurement Signals

Benchmarks / Datasets

Ts-Bench

Reported Metrics

f1

Research Brief

Deterministic synthesis

To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin. HFEPX signals include Automatic Metrics with confidence 0.45. Updated from current HFEPX corpus.

Generated Mar 13, 2026, 9:28 PM · Grounded in abstract + metadata only

Key Takeaways

  • To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin.
  • In parallel, we present Breeze Guard, an 8B safety model derived from Breeze 2, our previously released general-purpose Taiwanese Mandarin LLM with strong cultural grounding from…

Researcher Actions

  • Treat this as method context, then pivot to protocol-specific HFEPX hubs.
  • Cross-check benchmark overlap: Ts-Bench.
  • Validate metric comparability (f1).

Caveats

  • Generated from title, abstract, and extracted metadata only; full-paper implementation details are not parsed.
  • Low-signal flag detected: protocol relevance may be indirect.

Research Summary

Contribution Summary

  • To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin.
  • In parallel, we present Breeze Guard, an 8B safety model derived from Breeze 2, our previously released general-purpose Taiwanese Mandarin LLM with strong cultural grounding from its original pre-training corpus.
  • Empirically, Breeze Guard significantly outperforms the leading 8B general-purpose safety model, Granite Guardian 3.3, on TS-Bench (+0.17 overall F1), with particularly large gains in high-context categories such as scam (+0.66 F1) and…

Why It Matters For Eval

  • To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin.
  • In parallel, we present Breeze Guard, an 8B safety model derived from Breeze 2, our previously released general-purpose Taiwanese Mandarin LLM with strong cultural grounding from its original pre-training corpus.

Researcher Checklist

  • Gap: Human feedback protocol is explicit

    No explicit human feedback protocol detected.

  • Pass: Evaluation mode is explicit

    Detected: Automatic Metrics

  • Gap: Quality control reporting appears

    No calibration/adjudication/IAA control explicitly detected.

  • Pass: Benchmark or dataset anchors are present

    Detected: Ts-Bench

  • Pass: Metric reporting is present

    Detected: f1

Category-Adjacent Papers (Broader Context)

These papers are nearby in arXiv category and useful for broader context, but not necessarily protocol-matched to this paper.

Need human evaluators for your AI research? Scale annotation with expert AI Trainers.

Post a Job Get a Quote