Edge-FIT: Federated Instruction Tuning of Quantized LLMs

Discover Edge-FIT, a cutting-edge approach combining Federated Learning with IoT for efficient Edge AI solutions. This presentation delves into key innovations like PEFT integration, QLORA optimization, and resource minimization strategies, alongside experimental results comparing Edge-FIT to centralized training methods. Gain insights into its practical applications in IoT environments featuring Llama 2 and Phi-3-mini, and understand its benefits for privacy and resource efficiency. Finally,...