Fast Chain-of-Thought: A Glance of Future from Parallel Decoding Leads to Answers Faster
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910470128009216 |
|---|---|
| author | Zhang, Hongxuan Liu, Zhining Zhao, Yao Zheng, Jiaqi Zhuang, Chenyi Gu, Jinjie Chen, Guihai |
| author_facet | Zhang, Hongxuan Liu, Zhining Zhao, Yao Zheng, Jiaqi Zhuang, Chenyi Gu, Jinjie Chen, Guihai |
| contents | In this work, we propose FastCoT, a model-agnostic framework based on parallel decoding without any further training of an auxiliary model or modification to the LLM itself. FastCoT uses a size-varying context window whose size changes with position to conduct parallel decoding and auto-regressive decoding simultaneously, thus fully utilizing GPU computation resources. In FastCoT, the parallel decoding part provides the LLM with a quick glance of the future composed of approximate tokens, which could lead to faster answers compared to regular autoregressive decoding used by causal transformers. We also provide an implementation of parallel decoding within LLM, which supports KV-cache generation and batch processing. Through extensive experiments, we demonstrate that FastCoT saves inference time by nearly 20% with only a negligible performance drop compared to the regular approach. Additionally, we show that the context window size exhibits considerable robustness for different tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2311_08263 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Fast Chain-of-Thought: A Glance of Future from Parallel Decoding Leads to Answers Faster Zhang, Hongxuan Liu, Zhining Zhao, Yao Zheng, Jiaqi Zhuang, Chenyi Gu, Jinjie Chen, Guihai Computation and Language In this work, we propose FastCoT, a model-agnostic framework based on parallel decoding without any further training of an auxiliary model or modification to the LLM itself. FastCoT uses a size-varying context window whose size changes with position to conduct parallel decoding and auto-regressive decoding simultaneously, thus fully utilizing GPU computation resources. In FastCoT, the parallel decoding part provides the LLM with a quick glance of the future composed of approximate tokens, which could lead to faster answers compared to regular autoregressive decoding used by causal transformers. We also provide an implementation of parallel decoding within LLM, which supports KV-cache generation and batch processing. Through extensive experiments, we demonstrate that FastCoT saves inference time by nearly 20% with only a negligible performance drop compared to the regular approach. Additionally, we show that the context window size exhibits considerable robustness for different tasks. |
| title | Fast Chain-of-Thought: A Glance of Future from Parallel Decoding Leads to Answers Faster |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2311.08263 |