maybe we should make it * for both providers and ep. the quantization happens on pytorch model and the exported model is compatible with all eps.

their only requirement is that they require gpus to run. but that is a host machine requirement, not target provider or ep.

Does this supported_providers mean the output model supports which ep? Same concept for supported_accelerators?

yes. it's used by the auto-opt to filter out the passes based on the intended target ep

I see. I can see some other passes like QNNConversion is also in this list. But QNN model doesn't not have any ep concept. If this is only used by auto-opt, and auto-opt is targeting onnx model, should those unrelated pass whose output model is not onnx model be removed here?

QNN model does not have EP concept but you don't want to run QNNConversion pass if the user did not select QNN_EP.

@shaahji

Does this supported_providers mean the output model supports which ep? Same concept for supported_accelerators?

Yes, these fields dictate the target provider and accelerator for running the output model. We do not have a parallel support for host machine and depend on Pass.validate_config to make that decision.

I see. I can see some other passes like QNNConversion is also in this list. But QNN model doesn't not have any ep concept. If this is only used by auto-opt, and auto-opt is targeting onnx model, should those unrelated pass whose output model is not onnx model be removed here?

@shaahji what about this? i can see all passes needs to be registered here anyway?

Yes, all passes needs to be registered in this config since this is the only way to create a pass by importing modules.