Research map / Large language models / Language, safety and interpretability
An NLP Benchmark Dataset for Assessing Corporate Climate Policy Engagement
Gaku Morio, Christopher D Manning
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace BERT-base with ClimateBERT (climate-domain pretraining) | page-wise pipeline · Corporate climate policy engagement test (Document Q) | F-score: 63.5 -> 64.0 | Improved |
| replace ClimateBERT with RoBERTa-base | page-wise pipeline · Corporate climate policy engagement test (Document Q) | F-score: 64.0 -> 64.5 | Improved |
| replace RoBERTa-base with Longformer-base | page-wise pipeline · Corporate climate policy engagement test (Document S) | F-score: 53.1 -> 54.6 | Improved |
| scale Longformer-base to Longformer-large | page-wise pipeline · Corporate climate policy engagement test (Strict S) | F-score: 28.6 -> 31.5 | Improved |
| scale Longformer-base to Longformer-large | page-wise pipeline · Corporate climate policy engagement test (Document S) | F-score: 54.6 -> 57.3 | Improved |
| train a model instead of most-frequent baseline | Longformer-large · Corporate climate policy engagement test (Document P) | F-score: 46.7 -> 73.9 | Improved |
| use BERT-base instead of linear tf-idf model | page-wise pipeline · Corporate climate policy engagement test (Strict P) | F-score: 52.0 -> 50.2 | Hurt |
| use BERT-base instead of linear tf-idf model | page-wise pipeline · Corporate climate policy engagement test (Strict S) | F-score: 27.0 -> 25.8 | Hurt |
| use BERT-base instead of linear tf-idf model | page-wise pipeline · Corporate climate policy engagement test (Document Q) | F-score: 61.9 -> 63.5 | Improved |
| increase number of pages per document | Longformer-large · Corporate climate policy engagement test (Strict P) | F-score | Hurt |
| scale FlanT5-large to FlanT5-XL | comment generation · Comment generation test | ROUGE-1: 38.4 -> 39.5 | Improved |
| scale FlanT5-large to FlanT5-XL | comment generation · Comment generation validation | ROUGE-1: 43.4 -> 42.7 | Hurt |
Closest papers by meaning
- Assessing Large Language Models on Climate Information (ICML 2024 · similarity 0.84)
- ClimaQA: An Automated Evaluation Framework for Climate Question Answering Models (ICLR 2025 · similarity 0.82)
- InvestESG: A multi-agent reinforcement learning benchmark for studying climate investment as a social dilemma (ICLR 2025 · similarity 0.82)
- A Multimodal Benchmark for Framing of Oil & Gas Advertising and Potential Greenwashing Detection (NeurIPS 2025 · similarity 0.82)
- LLMCarbon: Modeling the End-to-End Carbon Footprint of Large Language Models (ICLR 2024 · similarity 0.81)
- Holistically Evaluating the Environmental Impact of Creating Language Models (ICLR 2025 · similarity 0.80)
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.
