Towards Scalable and Versatile Weight Space Learning
This work addresses the need for task-agnostic weight representations that can scale to larger models, which is incremental as it builds on hyper-representations but extends them to handle sequential processing and broader applications.
The paper tackles the problem of learning scalable and versatile representations of neural network weights, overcoming previous limitations in processing larger networks and task specificity, and demonstrates that SANE matches or exceeds state-of-the-art performance on benchmarks, particularly for initialization and larger ResNet architectures.
Learning representations of well-trained neural network models holds the promise to provide an understanding of the inner workings of those models. However, previous work has either faced limitations when processing larger networks or was task-specific to either discriminative or generative tasks. This paper introduces the SANE approach to weight-space learning. SANE overcomes previous limitations by learning task-agnostic representations of neural networks that are scalable to larger models of varying architectures and that show capabilities beyond a single task. Our method extends the idea of hyper-representations towards sequential processing of subsets of neural network weights, thus allowing one to embed larger neural networks as a set of tokens into the learned representation space. SANE reveals global model information from layer-wise embeddings, and it can sequentially generate unseen neural network models, which was unattainable with previous hyper-representation learning methods. Extensive empirical evaluation demonstrates that SANE matches or exceeds state-of-the-art performance on several weight representation learning benchmarks, particularly in initialization for new tasks and larger ResNet architectures.