4 ms·Self-Host Across GPU/CPUs Google's Gemma LLMs with a 2MB App3 points by 3Sophons 3y ago3Sophons 3y agoStill need room for the models themselves