Recycling Pretrained Checkpoints: Orthogonal Growth of Mixture-of-Experts for Efficient Large Language Model Pre-Training – Takara TLDR
Share Facebook Twitter LinkedIn Pinterest Email Funny Twitter spat between researchers arguing who was the first to invent an idea that has probably been around since 1990 😀 References: Links: YouTube: Twitter: BitChute: Minds: source
[Paper Analysis] On the Theoretical Limitations of Embedding-Based Retrieval (Warning: Rant)October 11, 2025