Delegate52: Long-Horizon Delegated Document Editing Across 52 Domains

DELEGATE52 is a benchmark dataset for evaluating large language models on long-horizon delegated document editing across 52 professional document domains. The dataset was developed by Microsoft to study the readiness of AI systems for delegated workflows, where knowledge workers instruct LLMs to edit documents on their behalf over long sessions. It was last updated on 2026-04-20.

Use Cases

• Benchmarking LLM performance on long-horizon editing tasks based on the described delegated workflow paradigm.
• Evaluating AI system readiness for professional document editing across diverse domains like crystallography files and music notation mentioned in the description.
• Studying new human-AI interaction paradigms where models are instructed to edit documents on a user's behalf.

Strengths

• Covers 52 distinct professional document domains, providing broad scope.
• Designed specifically for the novel research area of long-horizon delegated workflows.
• Developed by Microsoft, suggesting institutional backing.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Microsoft

Collection Method

Likely constructed for benchmarking purposes, but specific gathering method is not detailed.

Time Range

null

Freshness

Last updated 2026-04-20 00:31:06; freshness should be verified.

Geography

null