`包裹，长文用`
`/`
`划分小节。 3. 列表用`
- `，引用用`
`，代码用`
`，重要句子可用``。
4. 保留原文所有信息，不得增删改内容。
5. 不要输出``/``/``等文档级标签。
* 输入文本：一段关于“Looped Transformers / Loopback Transformer”的学术新闻/摘要。
2. 分析输入文本：
"Looped Transformers reduce parameter count by repeatedly applying shared blocks, but parameter sharing alone does not determine how the results of repeated computation are reused. We take the depth-memory view of residual connections and analyze which representation is stored after each block application and how later reads weight it. A full block output contains both its input and its new update, so weighting that output also reweights the history already carried by the input. Reading individual updates instead, with the shared stack's input kept on a separate path, gives separate controls over input strength, total update weight, and allocation across updates. The Loopback Transformer implements these controls: geometric decay sets the total update weight and supplies a recency prior, and attention learns how to allocate that weight. On ClimbMix, Loopback reaches lower validation loss than standard Transformer baselines with 40-75% of their Transformer parameters (excluding token embeddings and the output head) at reference scales from 191M to 3B, with training tokens and executed depth matched. It also has higher mean accuracy over thirteen downstream tasks at 1.4B and 3B. At 191M, 374M, and 778M, it has the lowest validation loss and the highest thirteen-task mean among the standard Transformer and seven looped residual designs."
*文本结构分析*：
- 第一句到第三句：介绍背景、提出视角（depth-memory view）以及分析全量输出（full block output）的问题。
- 第四句：提出解决方案（Reading individual updates instead...）。
- 第五句：具体实现（The Loopback Transformer implements...）。
- 第六句到第八句：实验结果（ClimbMix上的表现，参数量对比，下游任务准确率，不同规模下的对比）。
*划分小节（h2/h3）*：
可以将文本分为“方法分析”和“实验结果”两个部分，或者根据逻辑加上小标题。虽然规则说“长文用/
划分小节”，这段文本不算特别长，但加上小标题可以更好地体现语义化。
比如：
方法分析：深度记忆视角与循环计算
...
模型实现：Loopback Transformer
...
实验结果：性能与效率的突破
...
*检查规则4（不得增删改内容）*：
如果加了小标题，算不算“增删改内容”？通常增加小标题属于排版范畴，但为了严格遵守“不得增删改内容”，最好把原文的句子完整保留在``中，或者把小标题作为HTML标签的一部分，而不修改正文文字。
更安全的做法：不加额外的文字小标题，或者只使用纯标签结构。
如果必须用`
/
`，可以提炼原文的词句作为小标题，或者干脆只用`
`，因为规则说“长文用...”，这段其实也就是一段长文本。我可以直接把它分成几个`
`，或者用`
`，代码用``，重要句子可用``。 4. 保留原文所有信息，不得增删改内容。 5. 不要输出``/``/``等文档级标签。 * 输入文本：一段关于“Looped Transformers / Loopback Transformer”的学术新闻/摘要。 2. 分析输入文本： "Looped Transformers reduce parameter count by repeatedly applying shared blocks, but parameter sharing alone does not determine how the results of repeated computation are reused. We take the depth-memory view of residual connections and analyze which representation is stored after each block application and how later reads weight it. A full block output contains both its input and its new update, so weighting that output also reweights the history already carried by the input. Reading individual updates instead, with the shared stack's input kept on a separate path, gives separate controls over input strength, total update weight, and allocation across updates. The Loopback Transformer implements these controls: geometric decay sets the total update weight and supplies a recency prior, and attention learns how to allocate that weight. On ClimbMix, Loopback reaches lower validation loss than standard Transformer baselines with 40-75% of their Transformer parameters (excluding token embeddings and the output head) at reference scales from 191M to 3B, with training tokens and executed depth matched. It also has higher mean accuracy over thirteen downstream tasks at 1.4B and 3B. At 191M, 374M, and 778M, it has the lowest validation loss and the highest thirteen-task mean among the standard Transformer and seven looped residual designs." *文本结构分析*： - 第一句到第三句：介绍背景、提出视角（depth-memory view）以及分析全量输出（full block output）的问题。 - 第四句：提出解决方案（Reading individual updates instead...）。 - 第五句：具体实现（The Loopback Transformer implements...）。 - 第六句到第八句：实验结果（ClimbMix上的表现，参数量对比，下游任务准确率，不同规模下的对比）。 *划分小节（h2/h3）*： 可以将文本分为“方法分析”和“实验结果”两个部分，或者根据逻辑加上小标题。虽然规则说“长文用/
划分小节”，这段文本不算特别长，但加上小标题可以更好地体现语义化。 比如：
方法分析：深度记忆视角与循环计算
...
模型实现：Loopback Transformer
...
实验结果：性能与效率的突破
...
*检查规则4（不得增删改内容）*： 如果加了小标题，算不算“增删改内容”？通常增加小标题属于排版范畴，但为了严格遵守“不得增删改内容”，最好把原文的句子完整保留在``中，或者把小标题作为HTML标签的一部分，而不修改正文文字。 更安全的做法：不加额外的文字小标题，或者只使用纯标签结构。 如果必须用`
/
`，可以提炼原文的词句作为小标题，或者干脆只用`
`，因为规则说“长文用...”，这段其实也就是一段长文本。我可以直接把它分成几个`
`，或者用`
`概括。 让我们尝试排版为：
深度记忆视角与循环计算的重用机制
Looped Transformers reduce parameter count by repeatedly applying shared blocks, but parameter sharing alone does not determine how the results of repeated computation are reused. We take the depth-memory view of residual connections and analyze which representation is stored after each block application and how later reads weight it. A full block output contains both its input and its new update, so weighting that output also reweights the history already carried by the input.
独立读取更新与模型实现
Reading individual updates instead, with the shared stack's input kept on a separate path, gives separate controls over input strength, total update weight, and allocation across updates. The Loopback Transformer implements these controls: geometric decay sets the total update weight and supplies a recency prior, and attention learns how to allocate that weight.
实验结果与性能评估
On ClimbMix, Loopback reaches lower validation loss than standard Transformer baselines with 40-75% of their Transformer parameters (excluding token embeddings and the output head) at reference scales from 191M to 3B, with training tokens and executed depth matched. It also has higher mean accuracy over thirteen downstream tasks at 1.4B and 3B. At 191M, 374M, and 778M, it has the lowest validation loss and the highest thirteen-task mean among the standard Transformer and seven looped residual designs.
等等，规则4说“保留原文所有信息，不得增删改内容”。增加小标题（如“深度记忆视角...”）可能被视为“增加内容”。 如果不加小标题文字，只用 `` 包裹第一句话？不行，`
` 通常是标题。 如果整段就是一段话，我可以用 `
` 包裹，并用 `` 突出重要句子。 或者，我可以不加 `