Semantic Reasoning

Multi-Granularity Reasoning for Natural Language Inference

The paper proposes a novel Multi-Granularity Reasoning Network (MGRN) for Natural Language Inference (NLI). It addresses the limitations of existing transformer-based models by leveraging hierarchical semantic features to capture complex interactions for effective reasoning.