AI 中文总结
本研究针对Apple Silicon上的30亿参数智能体模型Nanbeige4.2-3B，修复其部署漏洞并提出分块预填充策略降低循环Transformer内存开销，使其在MCPMark等基准上完成智能体任务的比例提升至30%。
AI 中文摘要
Nanbeige4.2-3B是一款基于循环Transformer（Looped Transformer，LT）构建的30亿参数智能体模型，其通过复用一组层进行第二次前向传播，在不增加参数的前提下提升了有效深度。在Apple Silicon（MPS）上评估时，我们发现5个独立漏洞导致发布的检查点无法通过Hugging Face Transformers直接运行，包括RoPE缓冲区被静默置零、调用已移除的Transformers缓存API等。此外，我们发现修复这些漏洞仍不足以让模型完成智能体任务，原因是LT的层复用策略为实现参数效率，使峰值注意力内存翻倍。为此，我们提出分块预填充（chunked-prefill）策略，缓解了内存容量损耗，在32GiB共享内存上将允许的上下文宽度扩展了2.7倍。不过，即便内存开销降低，仍需补丁才能让Nanbeige4.2-3B可用；修复系统提示和MPS原生内存漏洞后，才能在标准MCP和工具调用基准上进行可靠评估。在MCPMark的子集上，修复后的模型完成高达30%的真实智能体任务（原模型为0%）；在BFCL上，单工具调用接近完美，但多数多工具测试失败。我们在该httpsURL发布了修复后的检查点、系统提示优化器和评估工具。
英文摘要
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.
Comments8 pages, 3 Tables
