An end-edge system using a Subtask Manager for adaptive offloading and a Conditioning Scale Estimator for feature-based pruning cuts multi-condition T2I latency by nearly 25% while raising average generation quality by 6%.
Multi-agent rl-based industrial aigc service offloading over wireless edge networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MM 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Accelerating Multi-Condition T2I Generation via Adaptive Condition Offloading and Pruning
An end-edge system using a Subtask Manager for adaptive offloading and a Conditioning Scale Estimator for feature-based pruning cuts multi-condition T2I latency by nearly 25% while raising average generation quality by 6%.