DeepSeek 发布 Gemma-4-12B 加速模型，本地性能领先

精选理由

DeepSeek 把 Gemma-4-12B 加速了，还带视觉，本地跑起来应该是同尺寸里最强的之一。

AI 摘要

DeepSeek 推出加速模型集合，其中 Gemma-4-12B 可能包含视觉能力。该模型在其 12B 参数量级别中，本地运行性能显著优于同类竞品。该集合未包含 Qwen 3.5，推测因 DeepSeek 未采用线性注意力机制。

原文 · Clement Delangue

https://t.co/PLL0hNPbdv

huggingface.co/collections/de… Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex Good guy DeepSeek gives us accelerated models The most interesting one here is Gemma4-12B, I presume vision included. Might be the best local model in its weight class now, by some margin Qwen 3.5 not included because DS[park] doesn't do linear attention I guess 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 10 👀 1899 📊 2 ⚡

查看原推