Depth induces an implicit low-rank bias in deep unconstrained feature models trained with unregularized multiclass cross-entropy, promoting softmax codes over neural collapse via more efficient norm propagation.
arXiv preprint arXiv:2206.05794 , year=
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
years
2026 6representative citing papers
LLM-driven evolutionary search discovers unsupervised UQ methods as Python programs that improve ROC-AUC by up to 6.7% over manual baselines on atomic claim verification across 9 datasets with OOD generalization.
Deep linear network theory derives logarithmic decay for cross-entropy loss under gap-growth conditions versus polynomial closure for Schatten-regularized structural energy under late-time KL tails, separating fitting from simplification; conditional reductions extend this to ReLU MLPs with fixed ac
Weight decay controls distinct learning regimes in grokking transformers on modular arithmetic, tracked by new cheap attention-based diagnostics with empirical critical value and exponent fits.
Weight decay slows progressive sharpening at the edge of stability, inducing damped oscillations in CNNs and a phase transition to sub-2/η sharpness in MLPs driven by parameter-sharpness gradient alignment, yielding more stable NTK dynamics.
A 120B sparse MoE model with 460 experts was trained on one 8-GPU node to loss 1.78 using reversible recurrence and state-preserving scaling from a 1.78B dense seed, with 5.93B active parameters.
citing papers explorer
-
The Implicit Bias of Depth: From Neural Collapse to Softmax Codes
Depth induces an implicit low-rank bias in deep unconstrained feature models trained with unregularized multiclass cross-entropy, promoting softmax codes over neural collapse via more efficient norm propagation.
-
Evolutionary Search for Automated Design of Uncertainty Quantification Methods
LLM-driven evolutionary search discovers unsupervised UQ methods as Python programs that improve ROC-AUC by up to 6.7% over manual baselines on atomic claim verification across 9 datasets with OOD generalization.
-
Deciphering Two Training Clocks in Grokking via Deep Linear Network Theory with Conditional ReLU Reduction
Deep linear network theory derives logarithmic decay for cross-entropy loss under gap-growth conditions versus polynomial closure for Schatten-regularized structural energy under late-time KL tails, separating fitting from simplification; conditional reductions extend this to ReLU MLPs with fixed ac
-
Reversible Foundations: Training a 120B Sparse MoE through State-Preserving Scaling
A 120B sparse MoE model with 460 experts was trained on one 8-GPU node to loss 1.78 using reversible recurrence and state-preserving scaling from a 1.78B dense seed, with 5.93B active parameters.