REFLEX: Rethinking MoE Inference as Refinement-Aware Compute Allocation in Diffusion Language Models
Xiang Xia, Cheng Yan, Yiming Zhang, Jiazheng Liu, Hongyu Zhang, Wuyang Zhang
cs.AI, cs.CL
Submitted: 2026-08-03
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Program Synthesis with Large Language Models
- LLaDA2.0: Scaling Up Diffusion Language Models to 100B
- Evaluating Large Language Models Trained on Code
- TIDE: Efficient and Lossless MoE Diffusion LLM Inference with I/O-aware Expert Offload
- Training Verifiers to Solve Math Word Problems
- dMoE: dLLMs with Learnable Block Experts
- Mixtral of Experts
- Seed Diffusion: A Large-Scale Diffusion Language Model with High-Speed Inference
- Expert Threshold Routing for Autoregressive Language Modeling with Dynamic Computation Allocation and Load Balancing
- Route Experts by Sequence, not by Token
- Dream 7B: Diffusion Large Language Models
- Expert-Choice Routing Enables Adaptive Computation in Diffusion Language Models
- LLaDA-MoE: A Sparse MoE Diffusion Language Model
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems