More Expressive Feedforward Layers: Part I. Token-Adaptive Mixing of Activations

arXiv:2605.26647 · cs.LG, cs.AI, stat.ML · Submitted 2026-05-26 · Read on arXiv

cs.LG, cs.AI, stat.ML

Submitted: 2026-05-26

Updated: 2026-08-28

Comments: 31 pages

Code: https://github.com/karpathy/nanoGPT

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Activation Function — 36× in this paper · explained in 1 episode
• Transformer — 18× in this paper · explained in 2 episodes
• Ablation — 14× in this paper · explained in 1 episode
• Computational overhead — 4× in this paper · explained in 2 episodes
• Mixture of Activations — 4× in this paper
• Trace — 4× in this paper · explained in 2 episodes
• Adaptivity — 3× in this paper · explained in 1 episode
• Ablation study — 2× in this paper · explained in 1 episode
• Large Language Models (LLMs) — 2× in this paper · explained in 6 episodes
• Scaling Laws — 2× in this paper · explained in 4 episodes
• Transformer Architecture — 2× in this paper · explained in 2 episodes
• Accuracy — 1× in this paper · explained in 1 episode
• Adam — 1× in this paper · explained in 2 episodes
• AdamW Optimizer — 1× in this paper · explained in 1 episode
• Approximation Bound — 1× in this paper · explained in 1 episode
• Associative Memory — 1× in this paper · explained in 2 episodes
• Benchmark — 1× in this paper · explained in 2 episodes
• Curse of Dimensionality — 1× in this paper · explained in 6 episodes
• Data Augmentation — 1× in this paper · explained in 7 episodes
• Embeddings — 1× in this paper · explained in 2 episodes
• Generalization — 1× in this paper · explained in 13 episodes
• Masked Autoencoder (MAE) — 1× in this paper · explained in 5 episodes
• Mixture-of-Experts (MoE) — 1× in this paper · explained in 7 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Sparsity — 1× in this paper · explained in 4 episodes

Sources

• Learning Activation Functions to Improve Deep Neural Networks
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Learning Activation Functions: A new paradigm for understanding Neural Networks
• Gaussian Error Linear Units (GELUs)
• Training Compute-Optimal Large Language Models
• MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies
• Adam: A Method for Stochastic Optimization
• Muon is Scalable for LLM Training
• KAN: Kolmogorov-Arnold Networks
• Decoupled Weight Decay Regularization
• Learning Combinations of Activation Functions
• Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering
• Mish: A Self Regularized Non-Monotonic Activation Function
• GLU Variants Improve Transformer
• Adaptive Blending Units: Trainable Activation Functions for Deep Neural Networks
• LLaMA: Open and Efficient Foundation Language Models
• Quick and (not so) Dirty: Unsupervised Selection of Justification Sentences for Multi-hop Question Answering
• Qwen2 Technical Report
• HellaSwag: Can a Machine Really Finish Your Sentence?
• ReLU$^2$ Wins: Discovering Efficient Activation Functions for Sparse LLMs

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks