SPARK conditions LLM edits in neural architecture search on specific functional factors to reduce entanglement, delivering 28.1x speedup and 22.9% relative OOD accuracy gain on CLRS-DFS.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Structured Progressive Knowledge Activation for LLM-Driven Neural Architecture Search
SPARK conditions LLM edits in neural architecture search on specific functional factors to reduce entanglement, delivering 28.1x speedup and 22.9% relative OOD accuracy gain on CLRS-DFS.