AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

ESG-Bench: Benchmarking Long-Context ESG Reports for Hallucination Mitigation

Siqi Sun, Ben Peng Wu, Mali Jin, Peizhen Bai, Hanpei Zhang, Xingyi Song

AAAI 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning generalization and delta compression

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 four-step CoT fine-tuning over plain SFT (ablation)Gemma-2-2B-it · ESG-Benchoverall accuracy (%): 63.33 -> 92.00Improved
add four-step CoT fine-tuning over plain SFT (ablation)Llama-3.2-3B-Instruct · ESG-Benchoverall accuracy (%): 90.67 -> 96.00Improved
add four-step CoT fine-tuning over plain SFT (ablation)Mistral-7B-Instruct-v0.3 · ESG-Benchoverall accuracy (%): 80.00 -> 90.00Improved
four-step instead of two-step CoT fine-tuning (ablation)Gemma-2-2B-it · ESG-Benchoverall accuracy (%): 72.67 -> 92.00Improved
add two-step CoT fine-tuning over plain SFT (ablation)Llama-3.2-3B-Instruct · ESG-Benchoverall accuracy (%): 90.67 -> 92.33Improved
supervised fine-tuning instead of no fine-tuning (ablation)Gemma-2-2B-it · ESG-Benchoverall accuracy (%): 71.00 -> 63.33Hurt
supervised fine-tuning instead of no fine-tuning (ablation)Llama-3.2-3B-Instruct · ESG-Benchoverall accuracy (%): 76.00 -> 90.67Improved
supervised fine-tuning instead of no fine-tuning (ablation)Gemma-2-2B-it · HaluEvaloverall accuracy (%): 90.00 -> 67.00Hurt
supervised fine-tuning instead of no fine-tuning (ablation)Gemma-2-2B-it · BioASQoverall accuracy (%): 89.75 -> 74.50Hurt
add four-step CoT fine-tuning over plain SFT (ablation)Gemma-2-2B-it · HaluEvaloverall accuracy (%): 67.00 -> 94.00Improved
add two-step CoT fine-tuning over plain SFT (ablation)Llama-3.2-3B-Instruct · BioASQoverall accuracy (%): 81.25 -> 81.00No clear effect
add two-step CoT fine-tuning over plain SFT (ablation)Gemma-2-2B-it · HaluEvaloverall accuracy (%): 67.00 -> 69.25No clear effect

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