Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910040022056960 |
|---|---|
| author | Lee, Jeongtae Jo, Minjung Jeong, Hyunjoon Park, Gunho Woo, Sunghyeon Kim, Joonghoon Kwon, Se Jung Lee, Dongsoo |
| author_facet | Lee, Jeongtae Jo, Minjung Jeong, Hyunjoon Park, Gunho Woo, Sunghyeon Kim, Joonghoon Kwon, Se Jung Lee, Dongsoo |
| contents | Speculative decoding accelerates large language model inference by proposing tokens with a lightweight draft model and selectively accepting them using a target model. This work introduces DropMatch, a novel approach that matches draft tokens to the predictive distribution of the target model via Monte Carlo dropout applied exclusively to the LM head, enabling sampling-based acceptance decisions. By generating multiple decoding paths, our method forms an empirical token distribution against which draft tokens are evaluated for consistency. This acceptance mechanism enables the model to adaptively control the size of decoding paths under an appropriate dropout probability, preventing substantial distortion of the target model predictive distribution. The proposed method operates in a training-free, data-free, and calibration-free manner, requires no architectural modification to pretrained models, and can be orthogonally integrated with a wide range of existing speculative decoding and inference acceleration techniques. Experiments across multiple benchmarks demonstrate that our approach increases acceptance length while maintaining competitive task performance, yielding inference speedups ranging from 1.09x to 1.33x over the standard baseline, and up to an additional 1.09x speedup when applied on top of EAGLE3. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_03333 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding Lee, Jeongtae Jo, Minjung Jeong, Hyunjoon Park, Gunho Woo, Sunghyeon Kim, Joonghoon Kwon, Se Jung Lee, Dongsoo Computation and Language Speculative decoding accelerates large language model inference by proposing tokens with a lightweight draft model and selectively accepting them using a target model. This work introduces DropMatch, a novel approach that matches draft tokens to the predictive distribution of the target model via Monte Carlo dropout applied exclusively to the LM head, enabling sampling-based acceptance decisions. By generating multiple decoding paths, our method forms an empirical token distribution against which draft tokens are evaluated for consistency. This acceptance mechanism enables the model to adaptively control the size of decoding paths under an appropriate dropout probability, preventing substantial distortion of the target model predictive distribution. The proposed method operates in a training-free, data-free, and calibration-free manner, requires no architectural modification to pretrained models, and can be orthogonally integrated with a wide range of existing speculative decoding and inference acceleration techniques. Experiments across multiple benchmarks demonstrate that our approach increases acceptance length while maintaining competitive task performance, yielding inference speedups ranging from 1.09x to 1.33x over the standard baseline, and up to an additional 1.09x speedup when applied on top of EAGLE3. |
| title | Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2603.03333 |