AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Multilingual Abusive Comment Detection at Scale for Indic Languages

Vikram Gupta, Sumegh Roychowdhury, Mithun Das, Somnath Banerjee, Punyajoy Saha, Binny Mathew, hastagiri prakash vanchinathan, Animesh Mukherjee

NeurIPS 2022 · Large language models › Language, safety and interpretability · approach: Pruning and circuits in LLMs · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Pretrain XLM-R on 5M social media comments (AbuseXLMR vs XLM-R)XLM-R · MACD Hindi monolingualF1-macro: 86.11 -> 87.93Improved
Pretrain XLM-R on social media comments (AbuseXLMR vs XLM-R)XLM-R · MACD Telugu monolingualF1-macro: 89.44 -> 91.37Improved
Unbalanced split (1:5 abusive:non-abusive) instead of randomAbuseXLMR · MACD HindiF1-macro: 88.52 -> 85.69Hurt
Unbalanced split (1:5) instead of randomAbuseXLMR · MACD Hindiaccuracy: 88.53 -> 93.05Improved
Chronological split instead of randomAbuseXLMR · MACD MalayalamF1-macro: 88.14 -> 81.82Hurt
Zero-shot cross-lingual transfer Hindi to Tamil instead of in-language trainingAbuseXLMR · MACD TamilF1-macro: 88.60 -> 86.70Hurt
Zero-shot cross-lingual transfer Hindi to Tamil instead of in-language trainingXLM-R · MACD TamilF1-macro: 87.87 -> 71.82Hurt
Few-shot with 100 samples: AbuseXLMR vs XLM-RXLM-R · MACD HindiF1-macro: 48.49 -> 77.05Improved
Few-shot 500 samples vs 5 samplesAbuseXLMR · MACD TeluguF1-macro: 42.15 -> 88.71Improved
Joint training on all languages instead of monolingual (ablation)mBERT · MACD HindiF1-macro: 84.31 -> 85.44Improved
Joint training on all languages instead of monolingual (ablation)XLM-R · MACD HindiF1-macro: 86.11 -> 85.26Hurt
Joint training on all languages instead of monolingual (ablation)XLM-R · MACD MalayalamF1-macro: 85.42 -> 85.87Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy