HybridSpec: Throughput-Aware Speculative Decoding for Hybrid GDN-MoE Models
cs.AI, cs.LG
Submitted: 2026-11-02
Updated: 2026-11-02
Terminology
Sources
- SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding
- Accelerating Large Language Model Decoding with Speculative Sampling
- DFlash: Block Diffusion for Flash Speculative Decoding
- FASER: Fine-Grained Phase Management for Speculative Decoding in Dynamic LLM Serving
- DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
- Training Verifiers to Solve Math Word Problems
- Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decoding
- TreeWY: Speculative Verification for Gated DeltaNet Hybrids
- P-EAGLE: Parallel-Drafting EAGLE with Scalable Training
- Kimi Linear: An Expressive, Efficient Attention Architecture
- Jamba: A Hybrid Transformer-Mamba Language Model
- D-cut: Adaptive Verification Depth Pruning for Batched Speculative Decoding
- MiniMax-01: Scaling Foundation Models with Lightning Attention
- Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding
- Utility-Driven Speculative Decoding for Mixture-of-Experts
- Twig: Parallel and Memory-Efficient Tree Speculative Decoding for Hybrid-Attention LLMs
- SpecLA: Efficient Speculative Decoding for Linear-Attention Models
- Less Experts, Faster Decoding: Cost-Aware Speculative Decoding for Mixture-of-Experts
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- 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
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents