Unable to generate engine file

[stephen-derosa]

I am running cuda 12.8, tensorrt 10.7.0 and torch 2.6.0. When i try generating the engine file with python tools/trt.py -n yolox-x -c weights/yolox_x.pth I get the following error:

2025-03-05 01:42:21.741 | INFO     | __main__:main:60 - loaded checkpoint done.
[03/05/2025-01:42:22] [TRT] [I] [MemUsageChange] Init CUDA: CPU +0, GPU +0, now: CPU 1210, GPU 6014 (MiB)
[03/05/2025-01:42:22] [TRT] [I] [MemUsageChange] Init builder kernel library: CPU +1, GPU +0, now: CPU 1413, GPU 6218 (MiB)
[03/05/2025-01:42:24] [TRT] [W] ITensor::setType(Float) was called on non I/O tensor: output_0. This will have no effect unless this tensor is marked as an output.
[03/05/2025-01:42:24] [TRT] [E] IBuilder::buildSerializedNetwork: Error Code 9: API Usage Error (Target GPU SM 87 is not supported by this TensorRT release.)
Saving torch model to  ./YOLOX_outputs/yolox_x/model_trt.pth
2025-03-05 01:42:24.098 | ERROR    | __main__:<module>:93 - An error has been caught in function '<module>', process 'MainProcess' (6468), thread 'MainThread' (281473344879872):
Traceback (most recent call last):

> File "/workspaces/aion-yolo/offline/aion-yolox/tools/trt.py", line 93, in <module>
    main()
    └ <function main at 0xfffefc354900>

  File "/opt/venv/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           │     │       └ {}
           │     └ ()
           └ <function main at 0xfffefc4cba60>

  File "/workspaces/aion-yolo/offline/aion-yolox/tools/trt.py", line 77, in main
    torch.save(model_trt.state_dict(), os.path.join(
    │     │    │         │             │  │    └ <function join at 0xffff9e314ae0>
    │     │    │         │             │  └ <module 'posixpath' (frozen)>
    │     │    │         │             └ <module 'os' (frozen)>
    │     │    │         └ <function Module.state_dict at 0xfffefe130fe0>
    │     │    └ TRTModule()
    │     └ <function save at 0xfffefe63f560>
    └ <module 'torch' from '/opt/venv/lib/python3.12/site-packages/torch/__init__.py'>

  File "/opt/venv/lib/python3.12/site-packages/torch/nn/modules/module.py", line 2222, in state_dict
    hook_result = hook(self, destination, prefix, local_metadata)
                  │    │     │            │       └ {'version': 1}
                  │    │     │            └ ''
                  │    │     └ OrderedDict()
                  │    └ TRTModule()
                  └ <function TRTModule._on_state_dict at 0xfffefc34dc60>
  File "/opt/venv/lib/python3.12/site-packages/torch2trt-0.5.0-py3.12.egg/torch2trt/trt_module.py", line 60, in _on_state_dict
    state_dict[prefix + "engine"] = bytearray(self.engine.serialize())
    │          │                              │    └ None
    │          │                              └ TRTModule()
    │          └ ''
    └ OrderedDict()

AttributeError: 'NoneType' object has no attribute 'serialize'

What is the best way to resolve? Thank you in advance.