Google DeepMind 推出 T5Gemma：全新 Encoder-Decoder 架構 Gemma 模型系列
Original: T5Gemma: A new collection of encoder-decoder Gemma models
Google DeepMind has announced a new open-source large language model series called "T5Gemma." This release represents another significant…
Google DeepMind 發表全新開源模型系列「T5Gemma」。不同於目前主流的 Decoder-only 架構，T5Gemma 採用經典的 Encoder-Decoder 架構，將 T5 的設計理念與 Gemma 的現代化技術結合。這為需要雙向上下文理解的任務（如翻譯、摘要、分類）提供了更高效且強大的新選擇。
Google DeepMind has announced a new open-source large language model series called "T5Gemma." This release represents another significant milestone for Google in the open-source AI model space. The core distinguishing feature of T5Gemma is its "Encoder-Decoder" architecture, which differs from the "Decoder-only" architecture that has dominated recent years (such as the GPT series, Llama series, and standard Gemma models).
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on Google DeepMind Blog →Summaries are AI-generated; the original article is authoritative.