AutoMix: Automatically Mixing Language Models
Paper
•
2310.12963
•
Published
•
14
Speculative
-
-
Large Language Model Cascades with Mixture of Thoughts Representations for Cost-efficient Reasoning
Paper • 2310.03094 • Published • 12 -
MatFormer: Nested Transformer for Elastic Inference
Paper • 2310.07707 • Published • 1 -
DistillSpec: Improving Speculative Decoding via Knowledge Distillation
Paper • 2310.08461 • Published • 1 -
DialCoT Meets PPO: Decomposing and Exploring Reasoning Paths in Smaller Language Models
Paper • 2310.05074 • Published • 1 -
Confident Adaptive Language Modeling
Paper • 2207.07061 • Published • 1 -
LLMCad: Fast and Scalable On-device Large Language Model Inference
Paper • 2309.04255 • Published • 1 -
SpecInfer: Accelerating Generative LLM Serving with Speculative Inference and Token Tree Verification
Paper • 2305.09781 • Published • 4 -
Future Lens: Anticipating Subsequent Tokens from a Single Hidden State
Paper • 2311.04897 • Published • 1 -
Depth-Adaptive Transformer
Paper • 1910.10073 • Published • 1 -
Fast Inference from Transformers via Speculative Decoding
Paper • 2211.17192 • Published • 4 -
Accelerating Large Language Model Decoding with Speculative Sampling
Paper • 2302.01318 • Published • 2 -
RecycleGPT: An Autoregressive Language Model with Recyclable Module
Paper • 2308.03421 • Published • 7 -
OrchestraLLM: Efficient Orchestration of Language Models for Dialogue State Tracking
Paper • 2311.09758 • Published • 1 -
Small Language Models Improve Giants by Rewriting Their Outputs
Paper • 2305.13514 • Published • 2 -
SortedNet, a Place for Every Network and Every Network in its Place: Towards a Generalized Solution for Training Many-in-One Neural Networks
Paper • 2309.00255 • Published • 1 -
Sorted LLaMA: Unlocking the Potential of Intermediate Layers of Large Language Models for Dynamic Inference Using Sorted Fine-Tuning (SoFT)
Paper • 2309.08968 • Published • 22 -
Fast and Robust Early-Exiting Framework for Autoregressive Language Models with Synchronized Parallel Decoding
Paper • 2310.05424 • Published • 1 -
Fast Chain-of-Thought: A Glance of Future from Parallel Decoding Leads to Answers Faster
Paper • 2311.08263 • Published • 15 -
BiTA: Bi-Directional Tuning for Lossless Acceleration in Large Language Models
Paper • 2401.12522 • Published • 11 -
Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding
Paper • 2401.07851 • Published • 1 -
APAR: LLMs Can Do Auto-Parallel Auto-Regressive Decoding
Paper • 2401.06761 • Published • 1 -
Cascade Speculative Drafting for Even Faster LLM Inference
Paper • 2312.11462 • Published • 8 -
Speculative Contrastive Decoding
Paper • 2311.08981 • Published • 2 -
Answering Unseen Questions With Smaller Language Models Using Rationale Generation and Dense Retrieval
Paper • 2308.04711 • Published • 1 -
Hydra: Sequentially-Dependent Draft Heads for Medusa Decoding
Paper • 2402.05109 • Published -
Sequoia: Scalable, Robust, and Hardware-aware Speculative Decoding
Paper • 2402.12374 • Published • 3 -
Online Speculative Decoding
Paper • 2310.07177 • Published • 1 -
Speculative Streaming: Fast LLM Inference without Auxiliary Models
Paper • 2402.11131 • Published • 41 -
Ouroboros: Speculative Decoding with Large Model Enhanced Drafting
Paper • 2402.13720 • Published • 5 -
Recurrent Drafter for Fast Speculative Decoding in Large Language Models
Paper • 2403.09919 • Published • 20 -
Better & Faster Large Language Models via Multi-token Prediction
Paper • 2404.19737 • Published • 73 -
Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge
Paper • 2405.00263 • Published • 14 -
TriForce: Lossless Acceleration of Long Sequence Generation with Hierarchical Speculative Decoding
Paper • 2404.11912 • Published • 16 -
Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting
Paper • 2404.18911 • Published • 29 -
Chimera: A Lossless Decoding Method for Accelerating Large Language Models Inference by Fusing all Tokens
Paper • 2402.15758 • Published -
REST: Retrieval-Based Speculative Decoding
Paper • 2311.08252 • Published -
Accelerating Production LLMs with Combined Token/Embedding Speculators
Paper • 2404.19124 • Published -
Parallel Decoding via Hidden Transfer for Lossless Large Language Model Acceleration
Paper • 2404.12022 • Published -
Speculative Decoding with Big Little Decoder
Paper • 2302.07863 • Published -
KV-Runahead: Scalable Causal LLM Inference by Parallel Key-Value Cache Generation
Paper • 2405.05329 • Published -
Accelerating Speculative Decoding using Dynamic Speculation Length
Paper • 2405.04304 • Published • 2 -
SDSAT: Accelerating LLM Inference through Speculative Decoding with Semantic Adaptive Tokens
Paper • 2403.18647 • Published -
On Speculative Decoding for Multimodal Large Language Models
Paper • 2404.08856 • Published • 13 -
You Only Cache Once: Decoder-Decoder Architectures for Language Models
Paper • 2405.05254 • Published • 9 -
Lossless Acceleration of Large Language Model via Adaptive N-gram Parallel Decoding
Paper • 2404.08698 • Published -
Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference
Paper • 2405.18628 • Published -
EMS-SD: Efficient Multi-sample Speculative Decoding for Accelerating Large Language Models
Paper • 2405.07542 • Published -
OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure
Paper • 2406.17276 • Published -
Speculative Decoding via Early-exiting for Faster LLM Inference with Thompson Sampling Control Mechanism
Paper • 2406.03853 • Published -
Optimizing Speculative Decoding for Serving Large Language Models Using Goodput
Paper • 2406.14066 • Published • 1 -
S2D: Sorted Speculative Decoding For More Efficient Deployment of Nested Large Language Models
Paper • 2407.01955 • Published -
Make Some Noise: Unlocking Language Model Parallel Inference Capability through Noisy Training
Paper • 2406.17404 • Published • 1 -
Adaptive Draft-Verification for Efficient Large Language Model Decoding
Paper • 2407.12021 • Published -
SEED: Accelerating Reasoning Tree Construction via Scheduled Speculative Decoding
Paper • 2406.18200 • Published -
Parallel Speculative Decoding with Adaptive Draft Length
Paper • 2408.11850 • Published -
Improving Multi-candidate Speculative Decoding
Paper • 2409.10644 • Published -
Learning Harmonized Representations for Speculative Sampling
Paper • 2408.15766 • Published -
Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recycling
Paper • 2408.08696 • Published -
PipeInfer: Accelerating LLM Inference using Asynchronous Pipelined Speculation
Paper • 2407.11798 • Published • 1