Entity · paper

Beyond Benchmarks: Exposing the Hidden Crisis in Bangla Hate Speech Detection

paperactiveprovisionalbeyond-benchmarks-exposing-the-hidden-crisis-in-bangla-hate-speech-detection-ce5d004b·1 events·first seen Jul 14, 2026

Aliases: Beyond Benchmarks: Exposing the Hidden Crisis in Bangla Hate Speech Detection

Co-occurring entities

BanglaBERT fastText

More like this (12)

hate speech detection From Self to Other: Evaluating Demographic Perspective-Taking in LLM Hate Speech Annotation UC Berkeley Measuring Hate Speech Corpus BanglaBERT Words Speak Louder Than Code: Investigating Cognitive Heuristics in LLM-Based Code Vulnerability Detection Ethical and Technical Limits of Deepfake Speech Datasets Massive Text Embedding Benchmark UTS at ELOQUENT 2026 Voight-Kampff: structural shifts in AI writing bypass state-of-the-art detectors ExploitBench DataGovBench Bias Benchmark for Question Answering CVE-Bench

Recent events (1)

4arXiv · cs.CL·Jul 14, 2026·source ↗

Study exposes generalization failures in Bangla hate speech detection models

A new arXiv paper evaluates six NLP architectures for Bangla hate speech detection, finding severe performance degradation when models trained on benchmark datasets are applied to real-world social media content. BanglaBERT drops from 91.4% F1 on benchmarks to 63.4% on implicit hate speech involving sarcasm and emojis. The study highlights a broader generalization crisis for low-resource language moderation systems and finds that emoji-aware preprocessing recovers up to 12% of lost performance.

Evaluation and Benchmarking AI Safety Research Beyond Benchmarks: Exposing the Hidden Crisis in Bangla Hate Speech Detection BanglaBERT fastText