{"slug": "what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in", "title": "What We Keep of What We Forget: Replacement Operators and Routing Semantics in Training-Free MoE Expert Pruning", "summary": "A study comparing replacement operators for training-free MoE expert pruning, holding the pruning mask fixed, evaluated three operator families on OLMoE, DeepSeek-V2-Lite and Mo (a third model whose name is truncated in the source). The work isolates the second of the two decisions training-free expert pruning makes — what a layer computes when its router still selects a removed expert — since published methods change the mask and the replacement operator at once. The comparison covers OLMoE, DeepSeek-V2-Lite and Mo, but the source excerpt does not report results.", "body_md": "Training-free expert pruning makes two decisions: which experts to remove, and what the layer computes when its router still selects a removed expert. Published methods change both at once. Holding the mask fixed, we compare replacement operators from three families on OLMoE, DeepSeek-V2-Lite and Mo", "url": "https://wpnews.pro/news/what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in", "canonical_source": "https://aiflash.com/news/145232/", "published_at": "2026-11-04 04:35:08+00:00", "updated_at": "2026-11-04 04:49:34.077581+00:00", "lang": "en", "topics": ["machine-learning", "large-language-models", "ai-research"], "entities": ["OLMoE", "DeepSeek-V2-Lite"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in", "markdown": "https://wpnews.pro/news/what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in.md", "text": "https://wpnews.pro/news/what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in.txt", "jsonld": "https://wpnews.pro/news/what-we-keep-of-what-we-forget-replacement-operators-and-routing-semantics-in.jsonld"}}