Green-LLM applies lexicographic multi-objective optimization to distribute LLM inference across renewable-powered heterogeneous data centers, jointly cutting carbon emissions and water consumption while keeping costs near minimum and latency under 2 seconds.
Litemoe: Customizing on- device llm serving via proxy submodel tuning,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NI 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Green-LLM: Optimal Workload Allocation for Environmentally-Aware Distributed Inference
Green-LLM applies lexicographic multi-objective optimization to distribute LLM inference across renewable-powered heterogeneous data centers, jointly cutting carbon emissions and water consumption while keeping costs near minimum and latency under 2 seconds.