MoE Model Routing Compared: DeepSeek Qwen Mixtral Llama 4 gpt-oss
Analysis details token routing differences across leading Mixture of Experts models.
Stories per day · 30 days
1 RSS
Relatedaction-graphs1agi1bgp1cascading1classification1efficiency1frr1language-models1llm-architecture1llm-classifiers1mirrormoe1mixture-of-experts1model-architecture1model-comparison1moe1olmo1omni-modal1query-cost1
29 stories, newest first · one card per story
Stories tagged #routing
Ubuntu fixes FRR RPKI regression in USN-7230-3
USN-7230-3 restores RPKI support in FRR after regression introduced by CVE-2024-55553 fix.
Route Anything framework improves AI system efficiency
Learned routers outperform fixed models by 0.3 points on 16 benchmarks while reducing policy time 24% and raising efficiency 8.82x on average.
LLM classifier circuits raise accuracy at a fixed query cost
On seven tasks spanning safety, health, and reasoning, the method beats routing and cascading baselines.
NIOCIS picks tool-use agent next hops without generating text
On 3,875 held-out security-operations decisions, a typed encoder matched the logged hop 32.2% of the time in about 10 ms.
Rethinking Routing Exploration in Modern Mixture-of-Experts
MirrorMoE fixes MoE routing under-exploration from shared experts by suppressing unneeded experts per token.
DAGFormer plans Transformer cross-layer connections with a topology predictor
On matched OLMo-2 pairs, WikiText-2 perplexity falls 17% to 25% versus a vanilla Transformer.
Collective Autonomous Vehicle Routing Needs Social Awareness for Coexistence with Human Drivers
Social marginal-cost rewards make autonomous-vehicle routing fairer to human drivers in mixed traffic.
Architect Launches Liquid Inference, a Real-Time Auction for LLM Inference
Architect launched Liquid Inference, an LLM router that auctions each prompt to the cheapest qualifying provider.
Sensitive-Topic Leakage Through LLM Routing Metadata: Measurement and Mitigation
Routing metadata from LLM gateways can reveal sensitive topics, including medical and self-harm requests, without content logs.
Distributionally Robust Mixture-of-Experts Training
DRMoET improves imperfect-routing robustness, raising a 10.3B MoE model's seven-task average from 0.6625 to 0.6767.
Stepped MoE: Segment-Level Routing with Configurable Inference Complexity
Stepped MoE combines elastic subnetworks with sparse routing so one model spans 1-4 billion parameters.
Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE
SplitMoE splits video-diffusion experts into semantic and generic roles, improving quality under a fixed parameter budget.
Do LLM Agents Execute the Plans They Declare? From Planning-Mode Declaration to Pattern-Specific Execution
Pattern-specific executors close LLM agents' plan-execution gap, lifting ALFWorld success from 0.48 to 0.92.
Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE
SplitMoE divides video-diffusion experts into semantic and generic roles, avoiding uniform routing and improving generation quality.
How to Loop MoE: Flatten the Experts, Untie the Attention
Foil flattens experts and unties attention in looped MoE models, lowering pretraining loss at equal parameters and compute.
How to Loop MoE: Flatten the Experts, Untie the Attention
Foil flattens experts and unties attention so looped mixture-of-experts models train with lower loss.
Routing Drift Alone Does Not Diagnose Failure in Merged MoE LLMs
Researchers find MoE routing drift after merging does not reliably indicate routing failure.
Routing Drift Alone Does Not Diagnose Failure in Merged MoE LLMs
Routing drift after MoE merging does not reliably indicate failure; source-route restoration often lacks task benefit.
MOPD-Router: Rethinking Teacher Routing in Multi-Teacher On-Policy Distillation
MOPD-Router routes multi-teacher distillation per token without domain labels, and ExpertAlign beats prior routing.
Jev in the Wild: A Data-Driven Analysis of the Jev Model's Functionality, Applications and Ecosystem
Study of 2,170 GitHub projects maps how the Jev decision model is used across domains.
SupersonicLabs/Julia-1 — new model trending #30 on Hugging Face
Supersonic Labs released Julia 1, a 144.3M-parameter model that picks among supplied answers for classification and routing.
3
MemoryAthena: Adaptive Routing over Latent and Generated Memories
MemoryAthena routes generated memory as a bounded correction to direct Engram retrieval.
COMPASS: Ordered Clustered Routing at 100K Scale
COMPASS combines search with learning-accelerated parallel sub-solvers for the Ordered Clustered TSP, scaling to 100K synthetic and 28.5K real e-commerce nodes.