# Nanbeige4.2-3B 在 Apple Silicon 上的修复：解决部署缺陷并降低 Looped Transformer 内存开销

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-08-14 08:00
- AIHOT 分数：49
- AIHOT 链接：https://aihot.virxact.com/items/cmsxjnmia02stroz0qi1v33wm
- 原文链接：https://arxiv.org/abs/2608.13987

## AI 摘要

Nanbeige4.2-3B 是 3B 参数智能体模型，采用 Looped Transformer 架构复用单层堆栈进行第二次前向传播以增加有效深度。研究者在 Apple Silicon（MPS）上发现五个独立 bug 阻碍其开箱运行，并引入 chunked-prefill 策略将可用上下文宽度扩展 2.7 倍。

## 正文

Nanbeige4.2-3B is a 3B-parameter agentic model built around a Looped Transformer (LT) that reuses one stack of layers for a second forward pass, adding effective depth without additional parameters. Evaluated on Apple Silicon (MPS), we identify five independent bugs which prevent the released checkpoint from running via Hugging Face transformers out of the box (including a silently-zeroed RoPE buffer and calls to removed transformers cache APIs). Furthermore, we show that fixing these bugs is still not sufficient for agentic tasks, due to the LT's layer-reuse strategy (which effectively doubles peak attention memory) used to achieve parameter efficiency. We thus introduce a chunked-prefill strategy which alleviates the incurred memory-capacity penalty, extending allowable context width by 2.7 times on 32~GiB shared memory. However, even with the reduced memory overhead, we show that patches are required to render Nanbeige4.2-3B usable; resolving both system prompt and MPS-native memory bugs finally allows reliable evaluation on standard MCP and tool-calling benchmarks. On a subset of MCPMark, the debugged model completes up to 30% of real agentic tasks (up from the original's 0%), while, on BFCL, it is near-perfect at single tool calls (yet fails the majority of multi-tool tests). We release the patched checkpoint, system prompt optimizer, and evaluation harnesses at https://github.com/johnhalloran321/Nanbeige4.2-3B-mps-fix.