Human Instructions
Cross-source consensus on Human Instructions from 1 sources and 5 claims.
1 sources · 5 claims
Uses
Risks & contraindications
Comparisons
Evidence quality
Where it comes from
Highlighted claims
- The Human Instructions evaluation set targets natural speech phenomena missing from standard benchmarks. — Building Interactive Real-Time Agents with Asynchronous I/O and Speculative Tool Calling
- The final Human Instructions set contains 177 valid sessions after human review. — Building Interactive Real-Time Agents with Asynchronous I/O and Speculative Tool Calling
- On Human Instructions, AsyncIO performed worse and slower than the synchronous SFT baseline. — Building Interactive Real-Time Agents with Asynchronous I/O and Speculative Tool Calling
- Naturalistic speech transcripts caused lower accuracy and increased AsyncIO latency due to repeated-action behavior. — Building Interactive Real-Time Agents with Asynchronous I/O and Speculative Tool Calling
- The Human Instructions results suggest synthetic streaming training does not fully cover spontaneous spoken interaction. — Building Interactive Real-Time Agents with Asynchronous I/O and Speculative Tool Calling