mlsanityData to 14 Oct 2026

PiERN: Token-Level Routing for Integrating High-Precision Computation and Reasoning

arXiv 2509.18169 ↗First posted 17 Sep 2025cs.LG cs.CE cs.CLcited by 0

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Tasks on complex systems require high-precision numerical computation to support decisions. However, current large language models (LLMs), even with enhanced reasoning capabilities, cannot integrate such computations as an intrinsic and interpretable capability with existing architectures. To this end, we propose Physically-isolated Experts Routing Network (PiERN), an architecture that directs computation and reasoning at token level, thereby enabling iterative alternation within a single chain of thought. We systematically evaluate PiERN on representative computation-reasoning tasks, including PDEBench and battery management tasks. Results show that PiERN achieves not only higher accuracy than directly finetuning LLMs but also significant improvements in response latency, token usage, GPU energy consumption, and experts routing accuracy compared with mainstream multi-agent approaches, while exhibiting no significant degradation in performance on MMLU and GLUE benchmarks. PiERN offers an efficient, interpretable, and scalable paradigm for interfacing language models with scientific systems.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 61 of its 72 references that are in the map. First posted on arXiv 17 Sep 2025.

Builds on

61 of 72 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024Counting Ability of Large Language Models and Impact of Tokenization25 Oct

Other references · 60

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2023GPT-4 Technical Reportcited by 18,396
  4. 2024The Llama 3 Herd of Modelscited by 14,715
  5. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  6. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  7. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  8. 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
  9. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  10. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,899
  11. 2024GPT-4o System Cardcited by 4,588
  12. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
38 more
  1. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,988
  2. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,601
  3. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,464
  4. 2023CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Societycited by 1,395
  5. 2023Gorilla: Large Language Model Connected with Massive APIscited by 1,332
  6. 2023HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Facecited by 1,190
  7. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,136
  8. 2024Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvementcited by 1,090
  9. 2024Chronos: Learning the Language of Time Seriescited by 759
  10. 2023CogVLM: Visual Expert for Pretrained Language Modelscited by 708
  11. 2023Time-LLM: Time Series Forecasting by Reprogramming Large Language Modelscited by 700
  12. 2023API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMscited by 591
  13. 2016Learning to Communicate with Deep Multi-Agent Reinforcement Learningcited by 575
  14. 2023Faith and Fate: Limits of Transformers on Compositionalitycited by 503
  15. 2025Search-o1: Agentic Search-Enhanced Large Reasoning Modelscited by 433
  16. 2022PDEBENCH: An Extensive Benchmark for Scientific Machine Learningcited by 413
  17. 2024MoE-LLaVA: Mixture of Experts for Large Vision-Language Modelscited by 300
  18. 2023UniControl: A Unified Diffusion Model for Controllable Visual Generation In the Wildcited by 200
  19. 2023Multimodal Large Language Models: A Surveycited by 195
  20. 2023Evaluating the Performance of Large Language Models on GAOKAO Benchmarkcited by 186
  21. 2023A Pretrainer's Guide to Training Data: Measuring the Effects of Data Age, Domain Coverage, Quality, & Toxicitycited by 182
  22. 2020Open Problems in Cooperative AIcited by 168
  23. 2023Understanding Catastrophic Forgetting in Language Models via Implicit Inferencecited by 144
  24. 2023Meta-Transformer: A Unified Framework for Multimodal Learningcited by 136
  25. 2023LLM4Drive: A Survey of Large Language Models for Autonomous Drivingcited by 136
  26. 2025WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agentcited by 119
  27. 2024Uni-MoE: Scaling Unified Multimodal LLMs with Mixture of Expertscited by 108
  28. 2024Revisiting Catastrophic Forgetting in Large Language Model Tuningcited by 99
  29. 2024Transformers Can Do Arithmetic with the Right Embeddingscited by 78
  30. 2022Limitations of Language Models in Arithmetic and Symbolic Inductioncited by 71
  31. 2015Probabilistic Numerics and Uncertainty in Computationscited by 60
  32. 2024CuMo: Scaling Multimodal LLM with Co-Upcycled Mixture-of-Expertscited by 56
  33. 2025A Survey on Large Language Models for Mathematical Reasoningcited by 52
  34. 2025Scaling Laws for Native Multimodal Modelscited by 50
  35. 2024MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detectioncited by 47
  36. 2024ChartMoE: Mixture of Diversely Aligned Expert Connector for Chart Understandingcited by 46
  37. 2024Evaluating Mathematical Reasoning of Large Language Models: A Focus on Error Identification and Correctioncited by 45
  38. 2024Optima: Optimizing Effectiveness and Efficiency for LLM-Based Multi-Agent Systemcited by 43

50 most cited of 60 All 60 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Atlas: Orchestrating Heterogeneous Models and Tools for Multi-Domain Complex Reasoning
  2. 2026ODAR: Principled Adaptive Routing for LLM Reasoning via Active Inference
  3. 2026Reasoning Is Not Free: Robust Adaptive Cost-Efficient Routing for LLM-as-a-Judge
  4. 2026TuringLLM: Efficiently Scaling Foundation Models Toward Physical AI
  5. 2026Orchestrating Intelligence: Confidence-Aware Routing for Efficient Multi-Agent Collaboration across Multi-Scale Models
  6. 2026Measured Joules, Learned Routes: Learning to Route for Energy-Efficient LLM Serving
  7. 2026Learning What to Retain: Gated-Memory Routing for Efficient Collaboration in Multi-Agent LLM Systems
  8. 2026Policy-Guided Stepwise Model Routing for Cost-Effective Reasoning