2026-09-25 · America/Los_Angeles · 社区动态 · #3

Ling Tiny 3.0 is a glimpse of the future

I've been playing around with Ling 3.0 Tiny, which is an 8 billion parameter model (MoE, 1B active). And I've had a lot of poignant thoughts as a result. Just for fun, I got it running with llama.cpp on an old laptop. This is a laptop from 2017 with a 7th gen i5 and 8 gigs of RAM, like barely even usable for modern tasks. No VRAM, no GPU. Well, I got Pi running on it and asked it to make a script to scan the local network for all available models on llama.cpp servers. It started chugging along at about 10 tokens per second. And 20 minutes later, it was done. It had several back and forth…

热度 48.0 / 100；排名与评分保留该期记录。