NoneType error when run torch.jit.script()

[YuantianGao]

Hello, I want to optimize the OFA subnet model but I meet the error:

Traceback (most recent call last):
  File "mobile_test.py", line 129, in <module>
    data_loader,bn_list[0],net_config))
  File "mobile_test.py", line 62, in prepare_subnet
    script_subnet = torch.jit.script(subnet)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_script.py", line 943, in script
    obj, torch.jit._recursive.infer_methods_to_compile
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_recursive.py", line 391, in create_script_module
    return create_script_module_impl(nn_module, concrete_type, stubs_fn)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_recursive.py", line 448, in create_script_module_impl
    script_module = torch.jit.RecursiveScriptModule._construct(cpp_module, init_fn)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_script.py", line 391, in _construct
    init_fn(script_module)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_recursive.py", line 428, in init_fn
    scripted = create_script_module_impl(orig_value, sub_concrete_type, stubs_fn)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_recursive.py", line 404, in create_script_module_impl
    property_stubs = get_property_stubs(nn_module)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/_recursive.py", line 697, in get_property_stubs
    properties_asts = get_class_properties(module_ty, self_name="RecursiveScriptModule")
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 153, in get_class_properties
    getter = get_jit_def(prop[1].fget, f"__{prop[0]}_getter", self_name=self_name)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 271, in get_jit_def
    return build_def(ctx, fn_def, type_line, def_name, self_name=self_name)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 305, in build_def
    build_stmts(ctx, body))
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 129, in build_stmts
    stmts = [build_stmt(ctx, s) for s in stmts]
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 129, in <listcomp>
    stmts = [build_stmt(ctx, s) for s in stmts]
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 279, in __call__
    return method(ctx, node)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 445, in build_Return
    return Return(r, None if stmt.value is None else build_expr(ctx, stmt.value))
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 279, in __call__
    return method(ctx, node)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 781, in build_Dict
    return DictLiteral(range, [build_expr(ctx, e) for e in expr.keys],
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 781, in <listcomp>
    return DictLiteral(range, [build_expr(ctx, e) for e in expr.keys],
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 278, in __call__
    raise UnsupportedNodeError(ctx, node)
  File "/home/yuantian/.local/lib/python3.6/site-packages/torch/jit/frontend.py", line 111, in __init__
    source_range = ctx.make_range(offending_node.lineno,
AttributeError: 'NoneType' object has no attribute 'lineno'

And this is my code:

ofa_net.set_active_subnet(ks=net_config['ks'], d=net_config['d'], e=net_config['e'])
subnet = ofa_net.get_active_subnet()
    
bn_mean = bn_list[0]
bn_var = bn_list[1]
calib_bn(subnet,data_loader,bn_mean,bn_var)
subnet = QuantizedModel(subnet)
#import torchvision
#subnet = torchvision.models.quantization.mobilenet_v3_small(pretrained=True)
#When I run torchvision's model. it can run successfully.
#fuse model
print(subnet)
fuse_model(subnet) 
    
#quantize model
subnet.qconfig = torch.quantization.default_qconfig
subnet.qconfig = torch.quantization.get_default_qconfig('qnnpack')
torch.backends.quantized.engine = 'qnnpack'
torch.quantization.prepare(subnet, inplace=True)
# Calibrate
print('Calibrating........................')
#evaluate_ofa_subnet(subnet,
	#imagenet_data_path, net_config, data_loader, batch_size=16)
    
#Convert model
torch.quantization.convert(subnet, inplace=True)
#optimize
script_subnet = torch.jit.script(subnet)
script_subnet_optimized = optimize_for_mobile(script_subnet)
return script_subnet_optimized

I can make sure there is None value in the model because when I run the torchvision's model, It can run successfully.
How to solve the problem?

And this is my code link:
https://drive.google.com/drive/folders/172QoQxYoB1fUPvwjudHWgBmRv-IfE8Mg?usp=sharing
You can run without the data loader in test1.py and mobile_test.py is the version that with data loader.
The database link:
https://drive.google.com/file/d/1LY0ERp-egZr7NxTw3bApu2H93s9CQuYV/view?usp=sharing.

Kind regards.
Thanks very much.

[Fr1ck]

Hi, thanks for your question. I meet the same error when using torch.jit.script(). Do you address it?