view article Article Welcome Gemma 3: Google's all new multimodal, multilingual, long context open LLM Mar 12 • 398
view post Post 3694 Introducing a new architecture, MedIT One – a single-token transformer with LSTM-like recurrence. It is extremely fast in training and inference, but we lack funding for large-scale training. Enjoy 🍓 https://github.com/MedITSolutionsKurman/medit-one See translation ❤️ 10 10 👍 5 5 + Reply