AdaptiveK Sparse Autoencoders: Dynamic Sparsity Allocation for Interpretable LLM Representations

Re-assign community