Skip to main content
PaperLLMsv1.0

Fast Inference from Transformers via Speculative Decoding

by Google Research · free · Last verified 2026-03-17

Introduced speculative decoding, a lossless inference acceleration technique that uses a smaller, faster draft model to propose multiple tokens, then verifies them in parallel with the target model in a single forward pass. This achieves 2-3x speedup without any degradation in output quality or distribution.

https://arxiv.org/abs/2211.17192
B+
B+Good
Adoption: AQuality: A+Freshness: B+Citations: AEngagement: F

Specifications

License
Open Access
Pricing
free
Capabilities
inference-acceleration, lossless-decoding, parallel-verification
Integrations
Use Cases
llm-inference-optimization, latency-reduction, production-serving
API Available
No
Tags
speculative-decoding, inference-efficiency, draft-model, lossless-acceleration, llm-inference
Added
2026-03-17
Completeness
100%

Index Score

74.7
Adoption
88
Quality
95
Freshness
74
Citations
82
Engagement
0

Put AI to work for your business

Deploy this paper alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.

Explore the full AI ecosystem on Agents as a Service