When Do Repetition Loops Lock? Truncation Sampling, Bounded Penalties, and a Blind Spot in DRY

arXiv:2610.24715v1 Announce Type: cross Abstract: Language models sometimes repeat verbatim until the generation limit. We study the exit rather than the entrance: given that a span has already been repeated n times, when is escape still possible? Along the loop the context is deterministic, so loop survival under any sampler can be computed exactly from one forward pass. Pure sampling leaves a positive escape probability at every step; top-p with threshold pi>1/2 locks a position exactly when the continuation probability reaches pi. A bounded penalty can leave a position locked; a penalty that outgrows the model's preference for the loop over unseen tokens guarantees escape under top-p. The same analysis predicts a blind spot in the widely deployed DRY sampler: its sequence breakers cap the penalty at lambda b^{g-1-a} for loops whose longest breaker-free run has g tokens. On eleven open models, 66% of locked units with g=6 stay locked under DRY and 1% with g=8. In free generation with llama.cpp, all 109 loops that survive DRY have gle 7. We propose DRY-C, which adds a term that grows exponentially with the number of verbatim recurrences, up to its look-back window. With it, none of 720 structured-output generations looped to the length limit, and task completion stayed close to DRY's.

Related

• Don't Repeat Yourself: Stopping Verbatim Loops at Sampling Time
• Self-Repulsive Sampling for Diffusion Language Models

Source: arXiv cs.LG | 2026-10-20