gemma4 推出12B dense model(可直接在aistudio中体验)
https://cdn3.ldstatic.com/original/4X/1/4/7/147f4a2a7152ed9b5e02ff9134f05b697496ca8c.pngGoogle – 3 Jun 26 (https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12B/)
https://cdn3.ldstatic.com/optimized/4X/5/7/6/57611695741e9e171eee2ca179c9651fe8b71ad2_2_690x388.jpeg
Introducing Gemma 4 12B: a unified, encoder-free multimodal model (https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12B/)
An overview of Gemma 4 12B, a model designed to bring high-performance multimodal intelligence directly to your laptop.
以下为官方公布的benchmark
https://cdn3.ldstatic.com/optimized/4X/8/9/9/899dafafacd69e6ed92b901d2a8380e5a8298a04_2_133x500.png
同样为多模态模型,采用encoder-free 架构训练,支持文字,图片,音频
可参阅相关技术报告
https://cdn3.ldstatic.com/original/4X/1/2/4/1244bc4f64448eac4e7834e0aacebbc3fae80a8a.png
developers.googleblog.com (https://developers.googleblog.com/gemma-4-12b-the-developer-guide/)
https://cdn3.ldstatic.com/optimized/4X/e/0/b/e0b63602c6d9d891fe857e394fc3337f6a1f34f7_2_690x345.jpeg
Gemma 4 12B: The Developer Guide- Google Developers Blog (https://developers.googleblog.com/gemma-4-12b-the-developer-guide/)
Meet Gemma 4 12B: the first medium-sized, encoder-free multimodal model capable of natively ingesting audio and video. Ideal for local AI development with 16GB VRAM, Hugging Face integrations, and drop-in local API servers.
采用sliding window attention技术 1024的滑动窗口大小,256k上下文长度.
谷歌blog介绍,其性能接近gemma4 26b model
https://cdn3.ldstatic.com/optimized/4X/7/7/4/774cd4e5a25e6eb2f360baf6297b993d3757e9e6_2_690x369.jpeg 进了星颖，你已经屏蔽掉抖音等平台更大的噪音，但星颖也有噪音，人多的地方就是这样。 有负反馈是好事， 停止做错的事情本身就是前进 。” 看到就转：生活有着无限的精彩，保持接纳一切的心态，就会巧遇无数个惊喜。告诉我们要以积极开放的心态去拥抱生活 找有结果的人付费学习， 花钱买经验，不要万事自己研究 。 今日提升，看到就是学到：提升数据就是提高用户动作的几率，而用户产生动作的诱因是什么？落脚点还是要放在内容上，放在信息的传递上。 围观，沙发在哪里？无脑送！
页:
[1]