AI Paper Atlas

Research map / Large language models / Alignment and preferences

Judging by the Rules: Compliance-Aligned Framework for Modern Slavery Statement Monitoring

Wenhao Xu, Akshatha Arodi, Jian-Yun Nie, Arsène Fansi Tchango

AAAI 2026 · Large language models › Alignment and preferences · approach: Steering and correcting LLM behavior

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
add CA-Judge rule-alignment reward on top of format and correctness rewards (ablation)CALLM (Qwen2.5-3B) · AIMS.au macro F1F1: 0.559 -> 0.639Improved
add CA-Judge reward (ablation)CALLM (Qwen2.5-3B) · AIMS.au C3 Risk DescriptionF1: 0.564 -> 0.712Improved
add CA-Judge reward (ablation)CALLM (Qwen2.5-3B) · AIMS.au C4 RemediationF1: 0.397 -> 0.570Improved
add CA-Judge reward (ablation)CALLM (Qwen2.5-3B) · AIMS.au C5 EffectivenessF1: 0.394 -> 0.439Mixed
add CA-Judge reward (ablation)CALLM (Qwen2.5-3B) · AIMS.au C2 StructureF1: 0.535 -> 0.572Improved
fine-tune with format and correctness rewards (GRPO) vs pre-trained base (ablation)Qwen2.5-3B-Instruct · AIMS.au macro F1F1: 0.293 -> 0.559Improved
add few-shot examples to CoT promptGPT-4o · AIMS.au macro F1F1: 0.559 -> 0.617Improved
CALLM vs GPT-4o on easy Approval criterionCALLM · AIMS.au ApprovalF1: 0.855 -> 0.786Hurt
train on AU, test cross-jurisdiction on UKCALLM · UK held-outmacro F1: 0.560 -> 0.620Improved
train on AU, test cross-jurisdiction on CanadaCALLM · Canada held-outmacro F1: 0.589 -> 0.617Improved
CA-Judge-trained CALLM vs GPT-4o few-shot on rule alignmentCALLM · AIMS.au testCA-Judge score: 0.53 -> 0.74Improved
raise GRPO KL weight beta from 0.001 to 0.04 (ablation)GRPO training · completion length during trainingreward hacking (long completions)Improved

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