resnet26 NPU 模型

import timm
import torch

model = timm.create_model("resnet26", pretrained=True, num_classes=1000)
model = model.to("npu:0")
model.eval()

# Run inference
with torch.no_grad():
    output = model(input_tensor)