---
id: 20261105-T0-13
title: "Dory1.0-4B：40亿参数循环模型，一个checkpoint适配不同算力预算"
title_en: "Dory1.0-4B: A 4B Looped Model That Adjusts Depth to Compute Budget"
url: https://ai.daily.yangsir.net/daily/20261105-T0-13
issue_date: 2026-11-05
publish_date: 2026-11-04T05:00:00.000Z
category: research
source_name: "arXiv cs.CL (NLP)"
source_url: https://arxiv.org/abs/2611.03998
---

# Dory1.0-4B：40亿参数循环模型，一个checkpoint适配不同算力预算

研究团队发布Dory1.0-4B，一个40亿参数的循环（潜在递归）语言模型，具备弹性深度：同一个checkpoint可服务不同的推理算力预算。模型基于归一化Transformer（nGPT）主干构建，层间采用交错结构。实际意义在于部署端只需维护一份权重，就能按场景在低算力和高算力之间切换。

## English Version

**Dory1.0-4B: A 4B Looped Model That Adjusts Depth to Compute Budget**

Researchers released Dory1.0-4B, a 4-billion-parameter looped (latent-recurrent) language model with elastic depth, so a single checkpoint serves different inference compute budgets. It is built on a normalized Transformer (nGPT) backbone with an interleaved layer structure. The practical payoff is deployment-side: one set of weights can shift between low- and high-compute settings depending on the scenario.

---

**来源**：[arXiv cs.CL (NLP)](https://arxiv.org/abs/2611.03998)

**详情页**：https://ai.daily.yangsir.net/daily/20261105-T0-13

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*