CausalBN-Bench: A Comprehensive Benchmark for Causal Learning Capability of LLMs

Keyphrases

• Learnability 100%
• Causal Learning 100%
• Large Language Models 100%
• Chain Structure 10%
• Learning Algorithm 10%
• Training Data 10%
• Large-scale Networks 10%
• Comprehensive Investigation 10%
• Data Distribution 10%
• Data Knowledge 10%
• Contextual Information 10%
• Collider 10%
• Lag behind 10%
• Information Distribution 10%
• Traditional Algorithm 10%
• Counterfactual Thinking 10%
• Causal Reasoning 10%
• Background Knowledge 10%
• Semantic Association 10%
• Distinct Entity 10%
• Model Texts 10%
• Text Comprehension 10%
• Closed Source 10%
• Chain-of-Thought 10%
• Causal Relationship 10%
• Reasoning Capabilities 10%
• Numerical Distribution 10%
• Causality Identification 10%
• Prompt Approaches 10%
• Causal Research 10%

Computer Science

• Large Language Model 100%
• Causal LLM 25%
• Underlying Data 12%
• Research Community 12%
• Contextual Information 12%
• Semantic Association 12%
• Background Knowledge 12%
• Comparative Evaluation 12%
• Training Data 12%
• Causal Relationship 12%
• Data Distribution 12%
• Learning Algorithm 12%
• Counterfactual Reasoning 12%
• Chain-of-Thought 12%