The authors developed TeleTune to address three challenges in learning from offline user telemetry: goals are not recorded, the environment cannot be replayed to evaluate updates, and logs may interleave several tasks without clear boundaries.
TeleTune learns agent skills from unfiltered user logs without goals or replay
The framework edits a textual skill library using action-prediction accuracy on fixed, interleaved telemetry, improving success rates by up to 12% over baselines.
Big Tech
Justin Chih-Yao Chen · Elias Stengel-Eskin · Yan Chen · Pol Llado · Scott Counts · Mohit Bansal · +3 more
UNC Chapel Hill · University of Texas at Austin · Microsoft
Research Digest··2 min read
The authors introduce TeleTune, a method for building a library of textual skills and workflows from offline telemetry logs that have no recorded goals, cannot be replayed, and may interleave multiple tasks.
Why this paper
From Microsoft and 2 others
In one line
TeleTune learns reusable textual skills from goal-free, non-replayable, interleaved telemetry logs by keeping only library edits that improve held-out action-prediction accuracy.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks (2 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§