SpecDrop: Parameter-Free Category-Conditioned Routing for Modular Specialization
Mixture-of-experts (MoE) networks pursue specialization through learned routers, gates, and load-balancing losses, yet at matched total-parameter budgets learned routers can underperform equal-weight No-Routing baselines. Is the bottleneck the routing algorithm, or the alignment between training-signal granularity and the target categories? We probe the question with SpecDrop, a fixed parameter-free routing scheme: each of $K$ branches receives weight $p_a$ for its assigned category and a small
Record details
Published: 4 August 2026
Source: arXiv cs.LG
Category: Research
Topics: Safety & alignment
Retrieved: 6 August 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Calibrating Trustworthiness: Co-Designing Metrics and Visualizations for Evaluating LLMs in Education
arXiv cs.HC · 4 August 2026
Latent Reward Registers for Diffusion Preference Alignment
arXiv cs.LG · 4 August 2026
Latent Reward Registers for Diffusion Preference Alignment
arXiv cs.LG · 4 August 2026
Socially Grounded Agentic AI: Coordinating Plural Perspectives through Social Theory
arXiv · 4 August 2026
When Privileged Guidance Misaligns: State-Matched Routing and Contextualized Self-Distillation for Multi-Turn Agents
HuggingFace Daily Papers · 4 August 2026
Agent Against Agent: An Agentic System for Automatic Prompt Injection Red Teaming
HuggingFace Daily Papers · 4 August 2026
How to cite this record
ethics.ai (4 August 2026), “SpecDrop: Parameter-Free Category-Conditioned Routing for Modular Specialization,” evidence record 16980, https://ethics.ai/record/16980 (originally published by arXiv cs.LG).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.