arXiv:2607. 03148v1 Announce Type: cross Abstract: Activation functions are considered an essential primitive for neural nonlinearity, i.
arXiv:2602. 12390v2 Announce Type: replace Abstract: We study neural networks with trainable low-degree rational activation functions and show that they are more expressive and parameter-efficient than modern piecewise-linear and smooth activations such as ELU, LeakyReLU, LogSigmoid, PReLU, ReLU, SELU, CELU, Sigmoid, SiLU, Mish, Softplus, Tanh, Softmin, Softmax, and LogSoftmax.
arXiv:2603. 06861v2 Announce Type: replace Abstract: Activation functions are fundamental to deep neural networks, governing gradient flow, optimization stability, and representational capacity.
arXiv:2602. 07518v3 Announce Type: replace-cross Abstract: Kolmogorov-Arnold Networks (KANs) shift neural computation from linear layers to learnable nonlinear edge functions, but implementing these nonlinearities efficiently in hardware remains an open challenge.
arXiv:2606. 00130v1 Announce Type: cross Abstract: We study Automatically Differentiable Nonlinear Tensor Networks (ADNTNs), a family of structured weight generators whose compact core tensors are trained end-to-end by reverse-mode automatic differentiation (AD).
arXiv:2602. 22352v2 Announce Type: replace-cross Abstract: With the continuous growth of neural network scales, low-precision quantization is widely used in edge accelerators.
arXiv:2606. 00130v2 Announce Type: replace-cross Abstract: Large deep neural networks are costly to store and deploy because inference must move and evaluate many parameters.
arXiv:2606. 16112v1 Announce Type: cross Abstract: Residual architectures are ubiquitous in deep learning, but they suffer from a subtle structural limitation: the norm of the residual stream can grow rapidly with depth.
arXiv:2609.36584v1 Announce Type: new Abstract: Analog in-memory computing (AIMC) offers an alternative for model training by executing matrix operations directly where weights are stored. However, s...
arXiv:2606. 14990v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) are standard tools for mechanistic interpretability, but current SAE families are constrained by fixed encoder nonlinearities such as ReLU, JumpReLU, and TopK.
arXiv:2607. 28418v1 Announce Type: cross Abstract: Pruning is a promising approach for improving the efficiency of LLMs.
arXiv:2605. 15435v2 Announce Type: replace Abstract: Standard deep-learning pipelines usually choose the network architecture before training and keep it fixed throughout optimization.
The paper introduces Mixture of Activations (MoA), a token‑adaptive feedforward network design that mixes multiple activation functions using lightweight gates while sharing linear projections. It also presents learnable activations (LA) as an input‑independent variant. The authors theoretically prove that MoA strictly surpasses both fixed‑activation FFNs and LA in expressive power, and empirically demonstrate that MoA achieves lower loss and better scaling on dense and MoE language models from 0.12 B to 2 B parameters with minimal overhead.