Crypto界消息,1 月 21 日,据量子位报道,DeepSeek-R1 发布一周年之际对新模型「MODEL1」进行了曝光。DeepSeek 在 GitHub 更新 FlashMLA 代码,横跨 114 个文件中有 28 处提到 MODEL1,与 V32 作为不同的模型出现。已知 V32 是 DeepSeek-V3.2,MODEL1 很可能是新的架构。代码中的具体差异体现在 KV 缓存布局、稀疏性处理和 FP8 解码方面,在内存优化上有多处不同。

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
0/400
No comments
  • Pin

Trade Crypto Anywhere Anytime
qrCode
Scan to download Gate App
Community
  • 简体中文
  • English
  • Tiếng Việt
  • 繁體中文
  • Español
  • Русский
  • Français (Afrique)
  • Português (Portugal)
  • Bahasa Indonesia
  • 日本語
  • بالعربية
  • Українська
  • Português (Brasil)