speculative decoding

Speculative Sampling via Exponential Races ACL 2025 Speculative Streaming: Efficient and Scalable Speculative Decoding with Multi-Stream Attention EMNLP 2025 Spec-VLA: Speculative Decoding for Vision-Language-Action Models with Relaxed Acceptance EMNLP 2025 SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning EMNLP 2025 Speed Without Sacrifice: Fine-Tuning Language Models with Medusa and Knowledge Distillation in Travel Applications ACL 2025 SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration ICLR 2025 TETRIS: Optimal Draft Token Selection for Batch Speculative Decoding ACL 2025 TokenSwift: Lossless Acceleration of Ultra Long Sequence Generation ICML 2025 Towards Optimal Multi-draft Speculative Decoding ICLR 2025 Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recycling ACL 2025 ZipAR: Parallel Autoregressive Image Generation through Spatial Locality ICML 2025 A Theoretical Perspective for Speculative Decoding Algorithm NeurIPS 2024 BASS: Batched Attention-optimized Speculative Sampling ACL 2024 Cascade Speculative Drafting for Even Faster LLM Inference NeurIPS 2024 DistillSpec: Improving Speculative Decoding via Knowledge Distillation ICLR 2024 Draft on the Fly: Adaptive Self-Speculative Decoding using Cosine Similarity EMNLP 2024 Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding ACL 2024 EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees EMNLP 2024 Generation Meets Verification: Accelerating Large Language Model Inference with Smart Parallel Auto-Correct Decoding ACL 2024 GliDe with a CaPE: A Low-Hassle Method to Accelerate Speculative Decoding ICML 2024 Graph-Structured Speculative Decoding ACL 2024 Guiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation ICML 2024 How Speculative Can Speculative Decoding Be? COLING 2024 Kangaroo: Lossless Self-Speculative Decoding for Accelerating LLMs via Double Early Exiting NeurIPS 2024 LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding ACL 2024