Dynamic Template Selection for Output Token Generation Optimization: MLP-Based and Transformer Approaches

TL;DR

This paper introduces Dynamic Template Selection (DTS), a cost-efficient method for adaptive response generation in large language models, using MLP and transformer approaches to match templates to query complexity, reducing token usage without sacrificing quality.

Contribution

It presents a novel formal framework for template selection, compares MLP and transformer routing methods, and demonstrates generalization across multiple LLM providers with extensive empirical validation.

Findings

MLP router achieves 90.5% accuracy, outperforming RoBERTa's 89.5%.

Token reduction ranges from 32.6% to 33.9% across providers.

Routing decisions generalize effectively across different LLMs.

Abstract

Contemporary large language model deployments typically employ uniform prompting strategies across diverse query types, applying verbose response patterns to both complex analytical tasks and straightforward factual questions. This one-size-fits-all methodology leads to substantial token inefficiency, a concern amplified by the significant cost differential between input and output tokens--the latter commanding 4-8x higher prices across major providers. We present Dynamic Template Selection (DTS), which adaptively matches response templates to query complexity, achieving significant cost reductions without compromising response quality. We compared two routing approaches: a simple MLP that uses pre-computed embeddings and a more complex fine-tuned RoBERTa transformer. Through comprehensive evaluation on 1,000 MMLU questions, we find that the MLP router achieves 90.5% routing accuracy on held-out test data, marginally exceeding RoBERTa's performance (89.5%) despite utilizing 125M fewer parameters. Notably, our empirical analysis reveals provider-agnostic behavior in template selection--routing decisions generalize effectively across 3 major LLM providers (OpenAI GPT-4, Google Gemini, and Anthropic Claude), as validated through 9,000 production API calls. While routing accuracy remains consistent at 90.5% across providers, observed token reductions vary from 32.6% to 33.9%, reflecting provider-specific generation characteristics. This work contributes several key elements: formal problem formulation with theoretical grounding in machine learning, four algorithms with corresponding complexity analyses, and extensive empirical validation across production systems.