What's your favorite medium to large (~50-150b params) MOE model? Aside from Qwen flash next.
英文标题:What's your favorite medium to large (~50-150b params) MOE model? Aside from Qwen flash next.
I recently got some more RAM on my desktop, which has opened up running larger MOEs by splitting them into RAM/VRAM. I know everyone loves Qwen flash next, but I'm curious what other ones are worth trying. On my list of to-try, I have: GPT OSS 120b (is this still usable nowadays?) Mistral 4 small
打开原文相关 5 · 机会 0
中文
I recently got some more RAM on my desktop, which has opened up running larger MOEs by splitting them into RAM/VRAM. I know everyone loves Qwen flash next, but I'm curious what other ones are worth trying. On my list of to-try, I have: GPT OSS 120b (is this still usable nowadays?) Mistral 4 small
中文全文翻译正在补齐,可先查看下方英文原文。
English / 原文
I recently got some more RAM on my desktop, which has opened up running larger MOEs by splitting them into RAM/VRAM. I know everyone loves Qwen flash next, but I'm curious what other ones are worth trying. On my list of to-try, I have: GPT OSS 120b (is this still usable nowadays?) Mistral 4 small