Failure-Transparent Agents: Benchmarking Post-Failure Reporting in Tool-Using Language Models
FTA is introduced, a controlled benchmark that fixes the failed observation and required evidence state before generation, making post-failure claims directly auditable and associated with substantially lower post-failure reporting errors while useful-response rates remain high within this blocked-task benchmark.
Jun-Ru Zhu, Shi-Ming Xie, Ai-Me-Lu-Fan Chen et al.
· 0 citations