It’ll spin up PyTorch properly to use DDP, so you can prepare the model that way if you want. Otherwise there’s a tutorial on distributed inference with accelerate here: Distributed Inference with 🤗 Accelerate
Hi, I’m having a similar problem but I could not apply the tutorial you’re referencing here because I’m working with Diffusers pipelines and I do not get the model isolated to use “prepare_pippy” as the tutorial says. Do you know if there is any possibility to isolate the model from the pipeline so I can follow that tutorial or should I use another approach?