Skip to content

Author

Chitraansh Pandey

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Jul 2026

Thermodynamic Weight Decay: Exploring Grokking Acceleration via Attention Specific Heat

Grokking -- the delayed generalization of neural networks long after they have memorized their training data -- wastes thousands of training epochs and is notoriously unpredictable. Building on the recent result that Transformer attention is formally isomorphic to a thermodynamic system, we treat the variance of attention logits as a specific heat Cv and show that its peak reliably precedes the generalization transition. We introduce CvAdamW, a drop-in AdamW variant that monitors Cv online and injects thermal energy by dynamically scaling weight decay when a phase transition is detected. Through a strictly iterative development process we identify three failure modes -- initialization noise, mini-batch micro-ripples, and slingshot blinding -- and resolve them with a memorization gate and an exponential-moving-average shock absorber. On modular arithmetic (a+b mod 97), CvAdamW enables grokking at epoch 2802 in a 4000-epoch budget where the baseline never groks. We further propose a scale-invariant z-score reformulation that removes task-specific hyperparameters, and evaluate it across 10 paired seeds. A paired analysis shows the cold-start variant reduces mean grokking latency by 257 epochs (6.0%; median 166 epochs; Wilcoxon p=0.049, Cohen's d=0.68, bootstrap 95% CI [53,489]), improving 8 of 10 seeds; on this single task Cv peaks before grokking in all 10 seeds. Our results indicate that neural networks may expose detectable precursors of impending generalization transitions, and that a physically motivated, proportional intervention can facilitate generalization within a fixed compute budget. Code and data are public.

Chitraansh Pandey · 1 citation

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.