Qwen 32B
A 32-billion-parameter large language model in Alibaba's Tongyi Qianwen (Qwen) series.
Timeline (last 90 days)
The launch of DeepSeek and the spread of models such as Qwen 32B have made it possible to run a local large model on a single consumer-grade graphics card.
A preliminary audit on the Qwen 32B model shows that about 90% of neurons have only one main meaning, about 7% have two related secondary meanings, and only 2-4% are uninterpreted.
All Facts (2)
The launch of DeepSeek and the spread of models such as Qwen 32B have made it possible to run a local large model on a single consumer-grade graphics card.
50%UnverifiedA preliminary audit on the Qwen 32B model shows that about 90% of neurons have only one main meaning, about 7% have two related secondary meanings, and only 2-4% are uninterpreted.
50%