Efficient Training of Sparse Autoencoders for Large Language Models via Layer Groups

Re-assign community