# LayerTrim offers training‑free pruning that keeps reasoning ability

> The new LayerTrim method trims MLP parts of decoder layers, speeding up Qwen3‑14B by 1.24× and cutting memory use by 19.7% while preserving long‑form reasoning.

Oossa · 2026-10-15 · https://oossa.com/en/layertrim-offers-training-free-pruning-that-keeps-reasoning-ability

Researchers introduced LayerTrim, a training‑free structured pruning technique for large language models. It removes redundant decoder layers but keeps attention mechanisms intact, which helps maintain performance on tasks that require long‑form reasoning. On the 14‑billion‑parameter Qwen3 model, LayerTrim makes prefill 1.24 times faster and reduces peak GPU memory by 19.7 percent compared with the unpruned model. Across three models and five benchmarks it improved average accuracy by up to 21.4 points over the prior RESP method.

## The facts

- Paper posted on Thu Oct 15 2026
- LayerTrim gives 1.24× faster prefill and 19.7% lower peak GPU memory on Qwen3‑14B

## Why it matters

Developers can deploy smaller, faster language models for reasoning tasks without costly retraining.

## Sources & references

1. [Unlocking Structured Pruning for Reasoning Language Models via Attention-Preserving MLP Trimming](https://arxiv.org/abs/2610.19121) – arXiv cs.LG, 2026-10-15

Last updated: 2026-10-15