HomeAI Tutorial

edge-cpu-inference

Public

Benchmarking 8 State-of-the-Art LLMs on commodity CPUs ($0.04/hr). Identified Qwen 2.5 (3B) as the Pareto-optimal solution for edge inference, outperforming DeepSeek R1 and Llama 2 in efficiency-accuracy trade-offs.

Creat2025-12-03T13:32:40
Update2025-12-05T15:42:13
0
Stars
0
Stars Increase

Related projects