LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation
🔗 https://huggingface.co/papers/2610.19975 ↗LatentLoop improves looped Mixture-of-Experts by compressing the first loop visit, reallocating saved computation to larger hidden dimensions, and using separate attention parameters, yielding lower validation loss and better benchmark scores.
內容為收集當日抓取的摘要原文。