Paper Detail

The Value of Thinking: Learning When to Reason in Hybrid Decision Models

Zhengxu Yu, Ke Xu, Meng Fang, Haitham Ammar, Weilin Luo, Jun Wang

huggingface Score 16.0

Published 2026-11-12 · First seen 2026-11-14

General AI

Abstract

Existing adaptive-reasoning methods shorten the explicit reasoning a model writes, or skip it by switching modes; either way the answer is still written out token by token. Rather than deciding how much to think, we explore a different paradigm, zero-or-T thinking: a single model either answers from the implicit thinking inside one forward pass, generating no token, or performs full autoregressive reasoning at a cost of T tokens when that is worth paying. We realise it in a hybrid decision model (HDM), where one frozen backbone serves both modes and a lightweight self-routing block chooses between them. The block estimates the value of thinking, the utility the slow mode is expected to add over the fast one, from what that implicit thinking has already produced, and the slow mode runs only where that value exceeds a price λ set per request. Where no per-decision label exists, as in episodic agentic tasks, we introduce Calibrated Decision Policy Optimisation (CDPO), a reinforcement-learning recipe that trains the block from episode-level returns. We evaluate HDM on MMLU-Pro, AGIEval, GPQA-Diamond, JevBench, BANKING77 and MiniWoB. Specifically, on JevBench HDM-4B scores 0.857 against Jev's 0.851 (+0.7%) at a median 0.179\,s against 0.652 (-72.5%). On the agentic venue MiniWoB HDM-27B cuts token cost by 80.1%.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{yu2026value,
  title = {The Value of Thinking: Learning When to Reason in Hybrid Decision Models},
  author = {Zhengxu Yu and Ke Xu and Meng Fang and Haitham Ammar and Weilin Luo and Jun Wang},
  year = {2026},
  abstract = {Existing adaptive-reasoning methods shorten the explicit reasoning a model writes, or skip it by switching modes; either way the answer is still written out token by token. Rather than deciding how much to think, we explore a different paradigm, zero-or-T thinking: a single model either answers from the implicit thinking inside one forward pass, generating no token, or performs full autoregressive reasoning at a cost of T tokens when that is worth paying. We realise it in a hybrid decision model},
  url = {https://huggingface.co/papers/2611.16961},
  keywords = {adaptive-reasoning, explicit reasoning, implicit thinking, zero-or-T thinking, hybrid decision model, frozen backbone, self-routing block, value of thinking, autoregressive reasoning, forward pass, Calibrated Decision Policy Optimisation, reinforcement learning, episode-level returns, per-decision label, episodic agentic tasks, slow mode, fast mode, code available, huggingface daily},
  eprint = {2611.16961},
  archiveprefix = {arXiv},
}

Metadata

{}