| Challenge: | Parallel scaling is a powerful paradigm to enhance reasoning capabilities in large language models. |
| Approach: | They propose a framework that enables efficient parallel scaling through dynamic pruning. |
| Outcome: | The proposed framework achieves token reductions of 65.73% to 88.50% compared to consensus sampling while maintaining competitive accuracy within 3.4 percentage points. |
Similar Papers
Hidden States as Early Signals: Step-level Trace Evaluation and Pruning for Efficient Test-Time Scaling (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to speed up parallel scaling have relied on similarity-based or confidence-based pruning, but these signals do not reliably indicate trace quality. |
| Approach: | They propose a pruning framework that evaluates reasoning steps using hidden states and dynamically prunes unpromising traces during generation. |
| Outcome: | The proposed framework reduces end-to-end inference latency by 45%–70% on average compared to self-consistency while improving reasoning accuracy. |
Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs (2026.findings-acl)
Copied to clipboard
Hongyuan Yuan, Xinran He, Run Shao, Bolei He, Xianwei Xue, Mengke Chen, Qiutong Pan, Haiwei Wang, Haifeng Li
| Challenge: | Extending CoT through RL can induce undesirable thinking patterns such as overthinking . prior work has focused on inefficient reflection, which manifests in two problematic patterns: Indiscriminate Reflection and Repetitive Reflectione . |
| Approach: | They propose a graph-based approach to optimize CoT by pruning each linear CoT into a directed acyclic graph with explicit dependency edges. |
| Outcome: | The proposed approach reduces the average reasoning tokens by 42% while maintaining or improving accuracy. |
Cut Your Losses! Learning to Prune Paths Early for Efficient Parallel Reasoning (2026.acl-long)
Copied to clipboard
| Challenge: | Parallel reasoning enhances Large Reasoning Models but incurs prohibitive costs due to futile paths caused by early errors. |
| Approach: | They propose a systematic taxonomy of path pruning to categorize methods by signal source and learnability. |
| Outcome: | The proposed model improves LRMs but incurs prohibitive costs due to futile paths caused by early errors. |
Slim-SC: Thought Pruning for Efficient Scaling with Self-Consistency (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies show that Test-Time Scaling (TTS) can improve reasoning performance without retraining the model. |
| Approach: | They propose a step-wise pruning strategy that identifies and removes redundant chains using inter-chain similarity at the thought level. |
| Outcome: | The proposed method reduces inference latency and KVC usage by up to 45% and 26% with R1-Distill while maintaining or improving accuracy. |
Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling (2026.findings-acl)
Copied to clipboard
Ivan Rodkin, Daniil Orel, Konstantin Smirnov, Arman Bolatov, Bilal Elbouardi, Besher Hassan, Yuri Kuratov, Aydar Bulatov, Preslav Nakov, Timothy Baldwin, Artem Shelmanov, Mikhail Burtsev
| Challenge: | Reasoning is a core capability of large language models, yet how multi-step reasoning is learned and executed remains unclear. |
| Approach: | They evaluate how large language models learn multi-step reasoning without memorization . they find that most neural architectures trained from scratch can learn rule inference . |
| Outcome: | The proposed framework fails to solve a natural-language proxy task with high accuracy. |
Think Clearly: Improving Reasoning via Redundant Token Pruning (2025.findings-emnlp)
Copied to clipboard
Daewon Choi, Jimin Lee, Jihoon Tack, Woomin Song, Saket Dingliwal, Sai Muralidhar Jayanthi, Bhavana Ganesh, Jinwoo Shin, Aram Galstyan, Sravan Babu Bodapati
| Challenge: | Recent large language models show promising capabilities in long-form reasoning . however, they tend to include substantial redundancy in reasoning paths . |
| Approach: | They propose a structure-aware pruning method that prioritizes removing redundant tokens . they remove redundant token and then resume the reasoning generation . |
| Outcome: | The proposed method shows strong performance on reasoning-intensive benchmarks without training. |
Pru-CoT: Towards Efficient Reasoning Distillation via Pruning Chain-of-Thought (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing heuristics fail to capture global causal logic due to rigid rules and limited search spaces. |
| Approach: | They propose a framework that extracts the essential logical structure from reasoning chains. |
| Outcome: | Experiments show that Pru-CoT models generate more compact reasoning paths compared to models trained on verbose data. |
Parallel Test-Time Scaling for Latent Reasoning Models (2026.acl-long)
Copied to clipboard
| Challenge: | Parallel test-time scaling is a pivotal approach for enhancing large language models. |
| Approach: | They propose two uncertainty-inspired stochastic strategies for parallel test-time scaling for latent reasoning models and a Latent Reward Model for aggregation. |
| Outcome: | The proposed model scales well with compute and enables effective trajectory selection. |
SwiftPrune: Hessian-Free Weight Pruning for Large Language Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a novel post-training pruning method relies on the Hessian matrix to perform pruning . current pruning methods are computationally intensive and lack performance due to second-order derivative calculations. |
| Approach: | They propose a Hessian-free weight pruning method that reduces computational burden . they use an Exponentially Weighted Moving Average technique to bypass weight sorting . |
| Outcome: | The proposed method achieves hardware-efficient model compression by eliminating computational intensive calculations. |
ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning (2026.acl-demo)
Copied to clipboard
Vladislav Smirnov, Quang-Chieu Nguyen, Sergey Senichev, Minh Ngoc Ta, Ekaterina Fadeeva, Artem Vazhentsev, Daria Galimzianova, Nikolai Rozanov, Viktor Mazanov, Jingwei Ni, Tianyi Wu, Igor Kiselev, Mrinmaya Sachan, Iryna Gurevych, Preslav Nakov, Timothy Baldwin, Artem Shelmanov
| Challenge: | Existing TTC scaling strategies and reasoning scorers are fragmented and evaluated under inconsistent protocols. |
| Approach: | They propose a framework for seamless test-time compute scaling of large language model reasoning . they use a modular Python library to implement state-of-the-art scaling strategy and scorer families . |
| Outcome: | The proposed framework evaluates performance and computational efficiency on mathematical and coding tasks. |