Search

1. colibri Hits GitHub Trending: Pure-C Engine Runs a 744B Model on Ordinary PCs

   

   …footprint of a 744B-parameter MoE model down to about 9.9 GB, so a computer with 8 GB of RAM can start such a …

   Flash·Oct 12

2. Nvidia-Backed Reflection Debuts Beam, Its First Open-Weight Frontier Model: 501B Params, 23B Active

   

   …pen-weight model: a text-only MoE with 501B total params and 23B active per token, pretrained on 23.8T tokens …

   Stories·Oct 6

3. Open-Source Strata Runs a 125-Billion-Parameter Model on a 12GB Gaming PC

   

   …tars in 12 days; the trick is MoE sparse activation plus tiered expert offloading, up to 94 tokens/s on an RTX…

   Stories·Oct 5