Pith. sign in

REVIEW 6 cited by

Learning to Route LLMs with Confidence Tokens

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.13284 v3 pith:IM7V7O23 submitted 2024-10-17 cs.CL cs.AIcs.LG

classification cs.CLcs.AIcs.LG
keywords confidencellmstokensanswersdownstreamlearningrouteself-ref
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Large language models (LLMs) have demonstrated impressive performance on several tasks and are increasingly deployed in real-world applications. However, especially in high-stakes settings, it becomes vital to know when the output of an LLM may be unreliable. Depending on whether an answer is trustworthy, a system can then choose to route the question to another expert, or otherwise fall back on a safe default behavior. In this work, we study the extent to which LLMs can reliably indicate confidence in their answers, and how this notion of confidence can translate into downstream accuracy gains. We propose Self-Reflection with Error-based Feedback (Self-REF), a lightweight training strategy to teach LLMs to express confidence in whether their answers are correct in a reliable manner. Self-REF introduces confidence tokens into the LLM, from which a confidence score can be extracted. Compared to conventional approaches such as verbalizing confidence and examining token probabilities, we demonstrate empirically that confidence tokens show significant improvements in downstream routing and rejection learning tasks.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Adaptive LLM Routing under Budget Constraints

    cs.LG 2025-08 conditional novelty 6.0 of 10

    LLM routing is framed as a budget-constrained contextual bandit, solved by a preference-prior initialized LinUCB variant with an online multi-choice knapsack cost policy.

  2. LAPO: Internalizing Reasoning Efficiency via Length-Adaptive Policy Optimization

    cs.AI 2025-07 conditional novelty 6.0 of 10

    LAPO is a two-stage RL method that reduces reasoning token usage by up to 40.9% while slightly improving accuracy, by learning per-problem length targets from the model's own correct answers and prompting the model to...

  3. Think in Blocks: Adaptive Reasoning from Direct Response to Deep Reasoning

    cs.AI 2025-08 unverdicted novelty 5.0 of 10

    A training framework that makes LLMs predict an integer reasoning budget as a number of blocks and adapt reasoning depth to task difficulty.

  4. Efficient Reasoning Through Suppression of Self-Affirmation Reflections in Large Reasoning Models

    cs.CL 2025-06 conditional novelty 5.0 of 10

    Suppressing low-probability 'wait' tokens that trigger self-affirmation reflections shortens reasoning-model output by 8-18% with roughly unchanged accuracy.

  5. CP-Router: An Uncertainty-Aware Router Between LLM and LRM

    cs.CL 2025-05 conditional novelty 5.0 of 10

    CP-Router uses conformal prediction set sizes from an LLM to decide whether to route a prompt to that LLM or to a more expensive reasoning model, cutting token use with minimal or no accuracy loss.

  6. Fine-Tuning and Prompt Engineering of LLMs, for the Creation of Multi-Agent AI for Addressing Sustainable Protein Production Challenges

    cs.AI 2025-06 conditional novelty 4.0 of 10

    A proof-of-concept multi-agent GPT system for microbial protein literature extraction shows both fine-tuning and prompt engineering improve cosine-similarity scores, with fine-tuning slightly ahead but more variable.

Pith tools