Tracing the Path to Grokking: Embeddings, Dropout, and Network Activation
This work addresses the challenge of understanding and predicting delayed generalization in neural networks, which is incremental as it builds on existing grokking research by providing new diagnostic tools.
The paper tackled the problem of predicting grokking, a phenomenon where neural networks generalize long after memorizing training data, by introducing practical metrics like dropout robustness and embedding similarity that forecast grokking behavior, with results showing these metrics correlate with transitions from memorization to generalization.
Grokking refers to delayed generalization in which the increase in test accuracy of a neural network occurs appreciably after the improvement in training accuracy This paper introduces several practical metrics including variance under dropout, robustness, embedding similarity, and sparsity measures, that can forecast grokking behavior. Specifically, the resilience of neural networks to noise during inference is estimated from a Dropout Robustness Curve (DRC) obtained from the variation of the accuracy with the dropout rate as the model transitions from memorization to generalization. The variance of the test accuracy under stochastic dropout across training checkpoints further exhibits a local maximum during the grokking. Additionally, the percentage of inactive neurons decreases during generalization, while the embeddings tend to a bimodal distribution independent of initialization that correlates with the observed cosine similarity patterns and dataset symmetries. These metrics additionally provide valuable insight into the origin and behaviour of grokking.