mlsanityData to 14 Oct 2026

Harnessing Chain-of-Thought Metadata for Task Routing and Adversarial Prompt Detection

arXiv 2503.21464 ↗First posted 27 Mar 2025cs.CL cs.AI cs.PFcited by 5

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

also in 1 more
Abstract (arXiv)

In this work, we propose a metric called Number of Thoughts (NofT) to determine the difficulty of tasks pre-prompting and support Large Language Models (LLMs) in production contexts. By setting thresholds based on the number of thoughts, this metric can discern the difficulty of prompts and support more effective prompt routing. A 2% decrease in latency is achieved when routing prompts from the MathInstruct dataset through quantized, distilled versions of Deepseek with 1.7 billion, 7 billion, and 14 billion parameters. Moreover, this metric can be used to detect adversarial prompts used in prompt injection attacks with high efficacy. The Number of Thoughts can inform a classifier that achieves 95% accuracy in adversarial prompt detection. Our experiments ad datasets used are available on our GitHub page: https://github.com/rymarinelli/Number_Of_Thoughts/tree/main.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 13 of its 34 references that are in the map. First posted on arXiv 27 Mar 2025.

Builds on

13 of 34 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Unveiling the Key Factors for Distilling Chain-of-Thought Reasoning25 Feb
  2. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Models21 May
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Models21 Mar

Other references · 10

  1. 2021On the Opportunities and Risks of Foundation Modelscited by 4,033
  2. 2024RouteLLM: Learning to Route LLMs with Preference Datacited by 682
  3. 2023MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuningcited by 487
  4. 2024Hybrid LLM: Cost-Efficient and Quality-Aware Query Routingcited by 372
  5. 2024Compressed Chain of Thought: Efficient Reasoning Through Dense Representationscited by 201
  6. 2023Large Language Model Routing with Benchmark Datasetscited by 172
  7. 2023Tree-Structured Parzen Estimator: Understanding Its Algorithm Components and Their Roles for Better Empirical Performancecited by 147
  8. 2022Holistic Evaluation of Language Modelscited by 131
  9. 2024An Early Categorization of Prompt Injection Attacks on Large Language Modelscited by 30
  10. 2024Privacy-preserved LLM Cascade via CoT-enhanced Policy Learningcited by 8

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 2.

5 papers.

  1. 2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Modelscited by 25
  2. 2025SynapseRoute: An Auto-Route Switching Framework on Dual-State Large Language Modelcited by 7
  3. 2026ESTAR: Early-Stopping Token-Aware Reasoning For Efficient Inferencecited by 3
  4. 2025Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLMscited by 2
  5. 2026The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Projectcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2023PromptRobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Prompts
  2. 2023Token-Level Adversarial Prompt Detection Based on Perplexity Measures and Contextual Information
  3. 2026RECAP: A Resource-Efficient Method for Adversarial Prompting in Large Language Models
  4. 2023PromptBench: A Unified Library for Evaluation of Large Language Models
  5. 2025PromptSleuth: Detecting Prompt Injection via Semantic Intent Invariance
  6. 2025OET: Optimization-based prompt injection Evaluation Toolkit
  7. 2026Scalable Prompt Routing via Fine-Grained Latent Task Discovery
  8. 2026Disentangling Adversarial Prompts: A Semantic-Graph Defense for Robust LLM Security