GaiaFlow combines semantic-guided diffusion tuning with early-exit and quantization methods to lower carbon emissions in neural information retrieval while maintaining competitive effectiveness.
Domain-specific question-answering systems: A case study of a carbon neutrality knowledge base.Sustainability, 17(5):2192
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
GaiaFlow: Semantic-Guided Diffusion Tuning for Carbon-Frugal Search
GaiaFlow combines semantic-guided diffusion tuning with early-exit and quantization methods to lower carbon emissions in neural information retrieval while maintaining competitive effectiveness.