Open-weight & owned stack · Slms ondevice
Bonsai Demonstrates Small-Model Edge Inference
A test reported that the 1.7B Bonsai model solved simple physics problems at about 9.1 tokens per second on a 12-watt Intel N97 system.
Read the original at reddit.comOpens the publisher's site in a new tab