D3D: Decoupled Diffusion Drafting for Asynchronous Speculative Decoding
In shortD3D removes target-model hidden state dependence in diffusion drafting, enabling asynchronous speculative decoding with up to 7.48× speedup.
Published 3 Nov 2026
10 Nov 2026
51 papers that day.
Picks that day
SLA: Scaling Spatial-Language-Action Model via Continual Pre-training
In shortSLA continually pre‑trains a vision‑language‑action model to align visual embeddings with 3D spatial representations, improving transferable spatial understanding for robots.
Published 5 Nov 2026
Everything else
49 more papers from this day-
Nov 5
SplatAnything: Gaussian Spacing Normalization for Feed-forward 3DGS across Arbitrary Scenes
TrendingOther ML
- Nov 2 P^2-MOPD: Accelerating Multi-Teacher On-Policy Distillation via Parameter Priors
- Oct 31 Decoupled Predictive Decision Making: A Shared Conditional Predictor for Agents
- Nov 4 LongBanana: An Expert-Verified Benchmark for Long-Context Multi-Reference Image Synthesis
- Nov 4 Confucius4-R2T2: A Stable-Prefix Streaming ASR Model for Real Real-Time Transcription
- Nov 5 PCBRouteAgent: How Close Are LLM Agents to Becoming PCB Routing Engineers?
- Nov 4 LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time
- Nov 5 Video-Coder: Reinforcing Agentic Coding for Video Generation
- Nov 4 EvoWorld: Towards Out-of-Sight World Evolution in Generative Video Models
- Nov 4 On the Opportunities and Challenges of Prefill-to-Autoregressive Cross-Model Transfer
- Nov 4 Beneath the Surface: What We Overlook When We Talk About Agent Evolution
- Nov 5 Terminal-style Mathematical Modeling Bench: The First Agentic and Verifiable Benchmark for Mathematical Modeling
- Nov 5 SAE Interprets MoE: Sparse AutoEncoders Reveal That MoE Experts Exhibit Human-Interpretable Fine-Grained Specialization
- Oct 30 Fast and Exact Relative-Position Attention via Structure-Aware Kernels
- Nov 3 LLMs Don't Know What They Know: Benchmarking Temporal Self-Knowledge for Financial Backtesting
- Nov 4 The Last Mile to the Real World: Agent-Native Primitives as Reusable Building Blocks for Solving Real-World Tasks
- Nov 3 Tools Need Not Speak in Text: Latent Context Compression for Agents
- Nov 4 SwarmEvolve: Latent Agent Swarms for Evolutionary Discovery and Optimization
- Nov 2 HeLM: Hierarchical Explicit Language Memory for Non-Markovian Manipulation
- Oct 24 AgentBaton: Efficient Model Collaboration in Agentic Tasks
- Nov 5 Exploring Weight-Space Programming: Capabilities, Generalization, and Limitations
- Nov 5 Muir: Linear Attention with Sensitivity-Allocated Gating and Matched-Filter Readout
- Nov 2 StepX-OPD: Rethinking Multi-Teacher On-Policy Distillation in Diffusion Models
- Nov 5 Steerable Multi-Objective Group-Relative Preference Optimization with Jointly Trained Low-Rank Adapters
- Nov 4 Signed but Steerable: Measuring and Closing the Text-Vector Integrity Gap in Offline Retrieval-Augmented Generation
- Nov 5 AutoCompress: Automated Hardware-Guided Model Compression
- Nov 6 Authorization Drift in LLM Agents: Evaluating Boundary Compliance in Multi-Step Tool Use
- Nov 5 CPU-ARC: Agentic Reasoning Caching for on-CPU Reasoning Reuse
- Nov 4 IWIBench: Evaluating Gradient-Free Model Driven Intra-weight Injection
- Nov 5 Frontier3DBench: Benchmarking Coding Agents for 3D Creation across Objects, Scenes, and Interactive Open Worlds
- Oct 30 Hierarchical Residuals: A Minimalist Decoupling of Residual Width from Compute Width
- Nov 3 Where to Connect? Boosting MLLMs via Dynamic Gated Pathways across ALL ViT and LLM Layers
- Nov 5 Towards Real-Time Full-Waveform LiDAR Transformers via Intensity-Guided Token Reduction and Physics-Aware Augmentation
- Nov 4 Reducing Representational Interference in Retrieval Embeddings with Trainable Prefix Tokens
- Oct 30 Learning to evolve by matching velocities
- Nov 3 Sketchpad: certified exact arithmetic with a small vision-language model
- Oct 26 OpenFrontiers: A Collection of Open Research Problems and a Benchmark for Scientific Rediscovery
- Nov 2 OneTrans-GSR: A Generative Slate Ranking Paradigm Unifying Reranking and Mix-Ranking with One Autoregressive Transformer
- Oct 31 "PRISM" Is Not All You Need: Cross-Topic Name Collisions and Scientific Paper Retrieval
- Oct 26 PREEN: Exemplar-Based Color Style Transfer with Smartphone Editor Presets
- Nov 5 Static Recovery Is Not Dynamic Stability: Dynamics-Aware Benchmarking of Protein Motif Scaffolding
- Oct 30 Hard Sample-Conditioned Decorrelated Attention for Robust Face Recognition
- Nov 2 Human-Centred, Risk-Constrained Evaluation of Financial Document Recognition: Evidence, Critical Errors, and Selective Automation
- Nov 1 ODE Policy: Robot Learning via Action Dynamics, Not Actions
- Nov 5 Toward the Era of Agentic Embodied Intelligence
- Nov 4 Information Games: Strategic Crowding and Firm Repositioning in Language-Model Space
- Nov 5 Knowing When to Teach: Learning from Teachable Prefixes in Multilingual Reasoning Distillation
- Nov 5 OmniLatent: Unified Latent for 3D Part Modeling
- Oct 28 SynthFaces: Balanced Large Scale Human Dataset