ComplexityNet: Increasing LLM Inference Efficiency by Learning Task Complexity
This work addresses efficiency optimization for LLM applications, particularly in resource-constrained environments, by introducing a novel method for task complexity assessment.
The paper tackled the problem of inefficient LLM inference by developing ComplexityNet, a model that predicts task complexity to allocate appropriate models, achieving 79% accuracy in complexity classification and reducing computational usage by 90% while maintaining 86.7% code generation accuracy.
We present ComplexityNet, a streamlined language model designed for assessing task complexity. This model predicts the likelihood of accurate output by various language models, each with different capabilities. Our initial application of ComplexityNet involves the Mostly Basic Python Problems (MBPP) dataset. We pioneered the creation of the first set of labels to define task complexity. ComplexityNet achieved a notable 79% accuracy in determining task complexity, a significant improvement over the 34% accuracy of the original, non fine-tuned model. Furthermore, ComplexityNet effectively reduces computational resource usage by 90% compared to using the highest complexity model, while maintaining a high code generation accuracy of 86.7%. This study demonstrates that fine-tuning smaller models to categorize tasks based on their complexity can lead to a more balanced trade-off between accuracy and efficiency in the use of Large Language Models. Our findings suggest a promising direction for optimizing LLM applications, especially in resource-constrained environments.