Exploring Test-time Scaling via Prediction Merging on Large-Scale Recommendation
This work addresses the underexplored area of test-time scaling for recommendation systems, offering a scaling-efficient approach that can be deployed online without affecting user-side inference time.
The paper tackles the problem of scaling deep learning recommendation systems at test time by generating diverse predictions through architectural heterogeneity and random initialization, achieving performance improvements that outperform parameter scaling under the same inference budget.
Inspired by the success of language models (LM), scaling up deep learning recommendation systems (DLRS) has become a recent trend in the community. All previous methods tend to scale up the model parameters during training time. However, how to efficiently utilize and scale up computational resources during test time remains underexplored, which can prove to be a scaling-efficient approach and bring orthogonal improvements in LM domains. The key point in applying test-time scaling to DLRS lies in effectively generating diverse yet meaningful outputs for the same instance. We propose two ways: One is to explore the heterogeneity of different model architectures. The other is to utilize the randomness of model initialization under a homogeneous architecture. The evaluation is conducted across eight models, including both classic and SOTA models, on three benchmarks. Sufficient evidence proves the effectiveness of both solutions. We further prove that under the same inference budget, test-time scaling can outperform parameter scaling. Our test-time scaling can also be seamlessly accelerated with the increase in parallel servers when deployed online, without affecting the inference time on the user side. Code is available.