Aggressive expert pruning in MoE LLMs extracts compact translation specialists that retain near-baseline quality after removing up to 75% of experts (or 90% with short SFT).
Shwai He, Daize Dong, Liang Ding, and Ang Li
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Extracting Small Translation Specialists from LLMs by Aggressively Pruning Experts
Aggressive expert pruning in MoE LLMs extracts compact translation specialists that retain near-baseline quality after removing up to 75% of experts (or 90% with short SFT).