Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU
AI Summary
The linked project, named Llama-modchip, describes running a 284-billion-parameter DeepSeek model using 64 GB of RAM without a GPU. The item is a technical computing project shared on Hacker News.
Article URL: https://github.com/synchopate/llama-modchip Comments URL: https://news.ycombinator.com/item?id=50411961 Points: 4 # Comments: 0
AI & Tech DeepSeek large language model Llama-modchip 64 GB RAM GPU-free inference open-source project