DECA partitions LLM parameters into blocks for sequential block-wise Adam optimization in decentralized non-IID settings to support efficient full-parameter fine-tuning.
Bullish” (positive) or “Bearish
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
-
DECA: Decentralizing Block-Wise Adam for Efficient LLM Full-Parameter Fine-Tuning on Non-IID Data
DECA partitions LLM parameters into blocks for sequential block-wise Adam optimization in decentralized non-IID settings to support efficient full-parameter fine-tuning.