Model Cascading for Code: A Cascaded Black-Box Multi-Model Framework for Cost-Efficient Code Completion with Self-Testing

TL;DR

This paper introduces a cascaded black-box framework combining model cascading and self-testing to optimize the cost-accuracy tradeoff in LLM-based code completion, reducing costs significantly while maintaining or improving accuracy.

Contribution

It proposes a novel, inference-time, black-box framework that dynamically balances model size and self-testing to optimize cost and accuracy in code generation.

Findings

Reduced average costs by 26% across models and datasets.

Achieved up to 70% cost reduction in best cases.

Maintained or improved accuracy compared to single-model schemes.

Abstract

The rapid advancement of large language models (LLMs) has significantly improved code completion tasks, yet the trade-off between accuracy and computational cost remains a critical challenge. While using larger models and incorporating inference-time self-testing algorithms can significantly improve output accuracy, they incur substantial computational expenses at the same time. Furthermore, servers in real-world scenarios usually have a dynamic preference on the cost-accuracy tradeoff, depending on the budget, bandwidth, the concurrent user volume, and users' sensitivity to wrong answers. In this work, we introduce a novel framework combining model cascading and inference-time self-feedback algorithms to find multiple near-optimal self-testing options on the cost-accuracy tradeoff in LLM-based code generation. Our approach leverages self-generated tests to both enhance accuracy and evaluate model cascading decisions. As a blackbox inference-time method, it requires no access to internal model parameters. We further propose a threshold-based algorithm to determine when to deploy larger models and a heuristic to optimize the number of solutions, test cases, and test lines generated per model, based on budget constraints. Experimental results show that our cascading approach reduces costs by an average of 26%, and up to 70% in the best case, across various model families and datasets, while maintaining or improving accuracy in natural language generation tasks compared to both random and optimal single-model self-testing schemes. To our knowledge, this is the first work to provide a series of choices for optimizing the cost-accuracy trade-off in LLM code generation with self-testing.