arxiv:2609.19671
Published on Sep 17
· Submitted by
Jaejun Shim on Sep 18
Upvote
45
Authors:
,
Abstract
Large Reasoning Models (LRMs) achieve strong performance on complex tasks but exhibit systematic inefficiency: they often overthink easy problems and underthink hard ones. Existing approaches based on uniform length penalties or rigid routing incur an efficiency tax, trading reduced computation on easy instances for accuracy loss on hard instances. We formulate efficient reasoning as an instance-adaptive computation allocation problem and propose When2Think, a post-training framework for hybrid reasoning that dynamically allocates computation based on problem difficulty. Our method introduces Instance-level Difficulty-Aware Control (IDAC), a reward-shaping mechanism that leverages pre-computed reference statistics (accuracy and token usage) to regulate reasoning depth. Combined with verifier-based rewards and batch-wise standardized advantages, IDAC enables stable critic-free optimization without learned reward models or online reference-model queries. When2Think encourages direct answering on easy instances while preserving extended reasoning on hard instances, thereby learning when to use System 1 (NoThink) versus System 2 (Think). Experiments on mathematical benchmarks demonstrate improved accuracy-efficiency trade-offs: on AIME24, Pass@3 increases by 10.0% while token usage is reduced by 27.9% relative to the base model, and on AIME25, When2Think achieves 40.0% Pass@3, outperforming compression and routing-only baselines.
View arXiv page View PDF Add to collection
Community
Paper author Paper submitter 5 days ago
Efficient Reasoning models by RFT
4 days ago
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
-
QLPO: Quadrant-weighted Sampling for Length-aware Policy Optimization (2026)
-
Off-Context GRPO: Learning to Reason on Hard Problems using Privileged Information (2026)
-
ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoning (2026)
-
SignalReasoner: Assessing the Upper Bound of 3B Models for Signal Mathematical Reasoning (2026)
-
AdaThinkV: Adaptive Thinking for Token-Efficient Video Reasoning (2026)
-
Eliciting Self-Verification in Multimodal Reasoning Agents with Reinforcement Learning (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on HF Mirror checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
· Sign up or log in to comment
Upvote
45
Models citing this paper 2
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2609.19671 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2609.19671 in a Space README.md to link it from this page.