LobRA reduces GPU seconds for multi-tenant LoRA fine-tuning by 45.03%-60.67% through heterogeneous FT replicas and per-step workload-balanced dispatching.
Smith, Luke Zettlemoyer, Wen-tau Yih, and Mike Lewis
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2025 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
LobRA: Multi-tenant Fine-tuning over Heterogeneous Data
LobRA reduces GPU seconds for multi-tenant LoRA fine-tuning by 45.03%-60.67% through heterogeneous FT replicas and per-step workload-balanced dispatching.