Integrated FP16 support in the Stable Diffusion model to reduce memory usage and potentially accelerate the training process.
Adjusted the model loading mechanism to evenly distribute the model across all available GPUs, facilitating balanced utilization and potentially improving performance.
Integrated FP16 support in the Stable Diffusion model to reduce memory usage and potentially accelerate the training process.
Adjusted the model loading mechanism to evenly distribute the model across all available GPUs, facilitating balanced utilization and potentially improving performance.