The var_name do not be updated in the cache when same metadat but different var name

[Max-ChenFei]

code = get_conversion("source_image", source_image_desc, "target_image")
the output is

target_image_desc)
import torch
var_2b2fdb1114854f7d8a1125e63e0724b6 = torch.from_numpy(source_image)
var_8e7a6666e26d4a46bf6c54428558934a = var_2b2fdb1114854f7d8a1125e63e0724b6.permute(2, 0, 1)
target_image = var_8e7a6666e26d4a46bf6c54428558934a.unsqueeze(0)

but
code = get_conversion("source_image", source_image_desc, "actual_image", target_image_desc)

still the same code, but the name of result should be actual_image here