Scaling refinement of natural language inference to naturalistic inputs is difficult because long, syntactically rich inputs and deep multi-step arguments amplify autoformalisation errors, where a single local mismatch can invalidate the proof.

Authors

• Person: Not available Organization: Hugging Face
  Daily Papers - Hugging Face

Sources

• Daily Papers - Hugging Face huggingface.co Hugging Face via serper

Referenced by nodes (1)

• natural language inference (NLI) concept