Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2506.08343 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915349929132032 |
|---|---|
| author | Wang, Chenlong Feng, Yuanning Chen, Dongping Chu, Zhaoyang Krishna, Ranjay Zhou, Tianyi |
| author_facet | Wang, Chenlong Feng, Yuanning Chen, Dongping Chu, Zhaoyang Krishna, Ranjay Zhou, Tianyi |
| contents | Recent advances in large reasoning models have enabled complex, step-by-step reasoning but often introduce significant overthinking, resulting in verbose and redundant outputs that hinder efficiency. In this study, we examine whether explicit self-reflection, signaled by tokens such as "Wait" and "Hmm", is necessary for advanced reasoning. We propose NoWait, a simple yet effective approach that disables explicit self-reflection by suppressing these tokens during inference. Extensive experiments on ten benchmarks across textual, visual, and video reasoning tasks show that NoWait reduces chain-of-thought trajectory length by up to 27%-51% in five R1-style model series, without compromising model utility. NoWait thus offers a plug-and-play solution for efficient and utility-preserving multimodal reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_08343 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency Wang, Chenlong Feng, Yuanning Chen, Dongping Chu, Zhaoyang Krishna, Ranjay Zhou, Tianyi Computation and Language Recent advances in large reasoning models have enabled complex, step-by-step reasoning but often introduce significant overthinking, resulting in verbose and redundant outputs that hinder efficiency. In this study, we examine whether explicit self-reflection, signaled by tokens such as "Wait" and "Hmm", is necessary for advanced reasoning. We propose NoWait, a simple yet effective approach that disables explicit self-reflection by suppressing these tokens during inference. Extensive experiments on ten benchmarks across textual, visual, and video reasoning tasks show that NoWait reduces chain-of-thought trajectory length by up to 27%-51% in five R1-style model series, without compromising model utility. NoWait thus offers a plug-and-play solution for efficient and utility-preserving multimodal reasoning. |
| title | Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2506.08343 |