ResearchBenchmarksInstruction Adherence and Security Robustness on StruQ Clean 1.0Follow84.87Capability ScoreText78.390880.072981.75583.4371Dec 3, 2025Evaluation ResultsMethodMethodLinksCapability ScoreASR (Naive)ASR (Ignore)ASR (Escape-S)ASR (Completion-R)ASR (Average)ASR (Worst)TextTraining Type=Vanilla...Training Type=Vanilla delimiter2025.1284.8732.2138.9420.6794.2346.5194.23DelimiterTraining Type=Special...Training Type=Special delimiter from StruQ2025.1284.2332.6947.623.5699.0450.7299.04CAHL2025.1283.623.0824.0420.1937.9826.3237.98ISE2025.1278.6421.1530.2920.6761.5433.4161.54