Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: Sim, Gihyeon
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2601.00894
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909980185067520
author Sim, Gihyeon
author_facet Sim, Gihyeon
contents Large language models apply uniform computation to all inputs, regardless of difficulty. We propose PonderTTT, a gating strategy using the TTT layer's self-supervised reconstruction loss to selectively trigger Test-Time Training (TTT) updates. The gating decision itself is training-free--requiring no learned classifier or auxiliary networks; only a single scalar threshold is initially calibrated on unlabeled data and continuously adapted via EMA to maintain target update rates. Our experiments with GPT-2 models (124M to 1.5B) on code language modeling (The Stack v2, teacher-forced perplexity) demonstrate that this signal is inference-compatible, requiring no ground-truth labels. Our Reconstruction Gating achieves 82-89% Oracle Recovery while being fully training-free, significantly outperforming Random Skip baselines (up to 16% lower loss on OOD languages).
format Preprint
id arxiv_https___arxiv_org_abs_2601_00894
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle When to Ponder: Adaptive Compute Allocation for Code Generation via Test-Time Training
Sim, Gihyeon
Machine Learning
Computation and Language
Large language models apply uniform computation to all inputs, regardless of difficulty. We propose PonderTTT, a gating strategy using the TTT layer's self-supervised reconstruction loss to selectively trigger Test-Time Training (TTT) updates. The gating decision itself is training-free--requiring no learned classifier or auxiliary networks; only a single scalar threshold is initially calibrated on unlabeled data and continuously adapted via EMA to maintain target update rates. Our experiments with GPT-2 models (124M to 1.5B) on code language modeling (The Stack v2, teacher-forced perplexity) demonstrate that this signal is inference-compatible, requiring no ground-truth labels. Our Reconstruction Gating achieves 82-89% Oracle Recovery while being fully training-free, significantly outperforming Random Skip baselines (up to 16% lower loss on OOD languages).
title When to Ponder: Adaptive Compute Allocation for Code Generation via Test-Time Training
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2601.00894