Prompt-Level Distillation: A Non-Parametric Alternative to Model Fine-Tuning for Efficient Reasoning

Hugging Face Daily Papers Papers

Summary

Prompt-Level Distillation (PLD) extracts reasoning patterns from teacher models into structured instructions for student model system prompts, improving performance on reasoning tasks without fine-tuning overhead.

Advanced reasoning typically requires Chain-of-Thought prompting, which is accurate but incurs prohibitive latency and substantial test-time inference costs. The standard alternative, fine-tuning smaller models, often sacrifices interpretability while introducing significant resource and operational overhead. To address these limitations, we introduce Prompt-Level Distillation (PLD). We extract explicit reasoning patterns from a Teacher model and organize them into a structured list of expressive instructions for the Student model's System Prompt. Evaluated using Gemma-3 4B, PLD improved Macro F1 scores on StereoSet (57\% to 90.0\%) and Contract-NLI (67\% to 83\%), while increasing LogiQA accuracy to 70\%. Similar results on Mistral Small 3.1 demonstrate cross-architecture generalizability, enabling these compact models to match frontier performance with negligible latency overhead. These expressive instructions render the decision-making process transparent, allowing for full human verification of logic, making this approach ideal for regulated industries such as law, finance, and content moderation, as well as high-volume use cases and edge devices.
Original Article
View Cached Full Text

Cached at: 06/16/26, 11:31 AM

Paper page - Prompt-Level Distillation: A Non-Parametric Alternative to Model Fine-Tuning for Efficient Reasoning

Source: https://huggingface.co/papers/2602.21103

Abstract

Prompt-Level Distillation extracts reasoning patterns from teacher models to enhance student model performance while maintaining interpretability and reducing latency.

Advanced reasoning typically requiresChain-of-Thought prompting, which is accurate but incurs prohibitive latency and substantial test-time inference costs. The standard alternative,fine-tuningsmaller models, often sacrifices interpretability while introducing significant resource and operational overhead. To address these limitations, we introducePrompt-Level Distillation(PLD). We extract explicit reasoning patterns from aTeacher modeland organize them into a structured list of expressive instructions for theStudent model’sSystem Prompt. Evaluated using Gemma-3 4B, PLD improvedMacro F1 scoreson StereoSet (57\% to 90.0\%) and Contract-NLI (67\% to 83\%), while increasingLogiQAaccuracy to 70\%. Similar results on Mistral Small 3.1 demonstratecross-architecture generalizability, enabling these compact models to match frontier performance with negligible latency overhead. These expressive instructions render the decision-making process transparent, allowing for full human verification of logic, making this approach ideal for regulated industries such as law, finance, and content moderation, as well as high-volume use cases and edge devices.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2602\.21103

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2602.21103 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2602.21103 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2602.21103 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Reasoning Compression with Mixed-Policy Distillation

arXiv cs.AI

This paper proposes Mixed-Policy Distillation (MPD), a framework that transfers concise reasoning behaviors from large teacher models to smaller student models, reducing token usage by up to 27.1% while improving performance.

Improving Reasoning Capabilities in Small Models through Mixture-of-Layers Distillation with Stepwise Attention on Key Information

arXiv cs.CL

This paper proposes a novel Chain-of-Thought distillation framework that transfers teacher models' stepwise attention on key information to student models through a Mixture-of-Layers module for dynamic layer alignment. The method achieves consistent performance improvements on mathematical and commonsense reasoning benchmarks by explicitly guiding student models to progressively focus on critical information during reasoning.

Adaptive Teacher Exposure for Self-Distillation in LLM Reasoning

Hugging Face Daily Papers

Adaptive Teacher Exposure for Self-Distillation (ATESD) improves LLM reasoning by dynamically adjusting how much of the reference reasoning the teacher shows the student during training, using a learnable policy controller and a discounted learning-progress reward. Experiments on math benchmarks show consistent improvements over existing self-distillation and RL baselines.