Open Issues Need Help
View All on GitHub🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support
AI Summary: Debug an AttributeError in the `accelerate` library's `remove_hook_from_module` function. The error occurs when using CPU offloading with LoRA weights in a text-to-image pipeline. The issue seems to be related to the order of operations: calling `enable_model_cpu_offload()` before loading and fusing LoRA weights causes the error.
🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support