onnxruntime bug - TensorRT does not natively support INT64

[mykeehu]

I can't solve a bug. I upload an image, check the Remove Background option, and get this error:

DepthMap v0.4.6 (7a3a7820)
creating background masks
2024-05-09 11:39:58.0160506 [W:onnxruntime:Default, tensorrt_execution_provider.h:83 onnxruntime::TensorrtLogger::log] [2024-05-09 09:39:58 WARNING] onnx2trt_utils.cpp:369: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
2024-05-09 11:40:30.8541291 [W:onnxruntime:Default, tensorrt_execution_provider.h:83 onnxruntime::TensorrtLogger::log] [2024-05-09 09:40:30 WARNING] The getMaxBatchSize() function should not be used with an engine built from a network created with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag. This function will always return 1.
2024-05-09 11:40:30.9487712 [W:onnxruntime:Default, tensorrt_execution_provider.h:83 onnxruntime::TensorrtLogger::log] [2024-05-09 09:40:30 WARNING] The getMaxBatchSize() function should not be used with an engine built from a network created with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag. This function will always return 1.

I've deleted the tensorrt and onnxruntime installations, but I still get this error after the specified directories are restored. How can I fix this error?

[mykeehu]

Ok, I found a solution, it seems!

• deleted the onnxruntime folders from the stable-diffusion-webui\venv\Lib\site-packages folder
• deleted stable-diffusion-webui\models\rem_bg folder
• went into venv and manually installed CUDA 12.1 with the following commands:

  pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
  pip install rembg[gpu,cli]

  Then the background removal worked like lightning.