Wiring Beats Blending: Structure-Aware Compensation for Transformer Downscaling

arXiv:2608.02829 · cs.LG, cs.AI, cs.CL · Submitted 2026-08-03 · Read on arXiv

cs.LG, cs.AI, cs.CL

Submitted: 2026-08-03

Updated: 2026-09-24

Comments: 15 pages, 5 figures. Independent research preprint

Code: https://github.com/rsdpyenugula/ScaleOp

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

Terminology

• Perplexity — 27× in this paper · explained in 3 episodes
• Pruning — 19× in this paper · explained in 4 episodes
• Distillation — 16× in this paper · explained in 2 episodes
• Transformer — 9× in this paper · explained in 2 episodes
• Knowledge Distillation — 6× in this paper · explained in 15 episodes
• Embeddings — 5× in this paper · explained in 2 episodes
• In International Conference on Learning Representations — 4× in this paper
• Ablation — 3× in this paper · explained in 1 episode
• Effective Rank — 3× in this paper · explained in 2 episodes
• Absorb — 2× in this paper · explained in 1 episode
• Consistency — 2× in this paper · explained in 3 episodes
• Core — 2× in this paper · explained in 2 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Regularization — 2× in this paper · explained in 3 episodes
• Trace — 2× in this paper · explained in 2 episodes
• Transformers — 2× in this paper · explained in 2 episodes
• Accuracy — 1× in this paper · explained in 1 episode
• Equivariance — 1× in this paper · explained in 3 episodes
• Fidelity — 1× in this paper · explained in 2 episodes
• Hyperparameters — 1× in this paper · explained in 2 episodes
• Interpretability — 1× in this paper · explained in 15 episodes
• Model Merging — 1× in this paper · explained in 3 episodes
• Representation Geometry — 1× in this paper · explained in 2 episodes
• Robustness — 1× in this paper · explained in 6 episodes
• Weight Decay — 1× in this paper · explained in 2 episodes

Sources

• Distilling the Knowledge in a Neural Network
• Compact Language Models via Pruning and Knowledge Distillation
• Weight subcloning: direct initialization of transformers using larger pretrained ones

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