claim
Current methods for natural language inference often handle failures via costly global regeneration because it is difficult to localise the responsible span or step from prover diagnostics.

Authors

Sources

Referenced by nodes (1)