inference speed
Inference speed refers to the time it takes for an AI model to make predictions once it has been trained. High inference speed is essential for applications requiring real-time decision-making, such as autonomous driving or online recommendations.
- CoVoMix2: Advancing Zero-Shot Dialogue Generation with Fully Non-Autoregressive Flow Matching
- Compressed and Smooth Latent Space for Text Diffusion Modeling
- FIGRDock: Fast Interaction-Guided Regression for Flexible Docking
- Fast Monte Carlo Tree Diffusion: 100× Speedup via Parallel and Sparse Planning
- Fast constrained sampling in pre-trained diffusion models
- FlexVAR: Flexible Visual Autoregressive Modeling without Residual Prediction
- Generalizable, real-time neural decoding with hybrid state-space models
- L-MTP: Leap Multi-Token Prediction Beyond Adjacent Context for Large Language Models
- LeMiCa: Lexicographic Minimax Path Caching for Efficient Diffusion-Based Video Generation
- Mamba Only Glances Once (MOGO): A Lightweight Framework for Efficient Video Action Detection
- MiniMax-Remover: Taming Bad Noise Helps Video Object Removal
- Optimize the Unseen - Fast NeRF Cleanup with Free Space Prior
- Pan-LUT: Efficient Pan-sharpening via Learnable Look-Up Tables
- RetrievalAttention: Accelerating Long-Context LLM Inference via Vector Retrieval
- ScaleDiff: Higher-Resolution Image Synthesis via Efficient and Model-Agnostic Diffusion
- Succeed or Learn Slowly: Sample Efficient Off-Policy Reinforcement Learning for Mobile App Control
- Test-Time Spectrum-Aware Latent Steering for Zero-Shot Generalization in Vision-Language Models
- Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator