# AI and # localhost and uh # DIY # offline # Data Privacy - Dell Laptop (16 GB RAM, NVIDIA Quadro P1000 with 4 GB VRAM) - LM Studio with gemma-4-e2b-it-qat - with
A user details their experience running the gemma-4-e2b-it-qat model locally on a Dell laptop with 16GB RAM and an NVIDIA Quadro P1000 GPU. They fed the model an 8MB text file extracted from eight years of Telegram chats, with the goal of having the AI respond as if it were the user. The user found that the AI did not fully comprehend the context, as demonstrated by its responses to questions about a proposed surf park in Krefeld. AI
IMPACT Demonstrates the capabilities and limitations of running smaller, open-source models locally for personalized AI interactions.