“Error Code 2: Internal Error (Assertion y->type() == signature.operandType(1) failed.” failure of TensorRT 8.5.2 when running sam_decoder.onnx on GPU RTX 3060

Description

I tried to transfer onnx file of sam to tensorRT for deployment, but it fails with the error below .

"""
[2023-05-15 15:36:25][info][trt_builder.cpp:474]:Compile FP32 Onnx Model 'D:/xiaowen/python/logs/sam/sam_vit_b_encoder.onnx'.
[2023-05-15 15:36:35][warn][trt_builder.cpp:33]:NVInfer: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See CUDA_MODULE_LOADING in https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars
[2023-05-15 15:36:36][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:736: While parsing node number 297 [Mod -> "onnx::Sub_211"]:
[2023-05-15 15:36:36][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:737: --- Begin node ---
[2023-05-15 15:36:36][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:738: input: "onnx::Mod_203"
input: "onnx::Mod_210"
output: "onnx::Sub_211"
name: "Mod_297"
op_type: "Mod"
attribute {
name: "fmod"
i: 0
type: INT
}

[2023-05-15 15:36:36][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:739: --- End node ---
[2023-05-15 15:36:36][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:742: ERROR: TensorRT\onnx_parser\builtin_op_importers.cpp:4742 In function importFallbackPluginImporter:
[8] Assertion failed: creator && "Plugin not found, are the plugin name, version, and namespace correct?"
[2023-05-15 15:36:36][error][trt_builder.cpp:519]:Can not parse OnnX file: D:/xiaowen/python/logs/sam/sam_vit_b_encoder.onnx
[2023-05-15 15:36:36][info][direct_sam.cpp:40]:Compile done
[2023-05-15 15:36:36][info][trt_builder.cpp:474]:Compile FP32 Onnx Model 'D:/xiaowen/python/logs/sam/sam_vit_b_decoder.onnx'.
[2023-05-15 15:36:38][warn][trt_builder.cpp:33]:NVInfer: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See CUDA_MODULE_LOADING in https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars
[2023-05-15 15:36:40][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:534: Found unsupported shape-tensor producing layer:Reciprocal_1095
[2023-05-15 15:36:40][warn][trt_builder.cpp:33]:NVInfer: Tensor DataType is determined at build time for tensors not marked as input or output.
[2023-05-15 15:36:40][warn][trt_builder.cpp:33]:NVInfer: Tensor DataType is determined at build time for tensors not marked as input or output.
[2023-05-15 15:36:40][error][trt_builder.cpp:30]:NVInfer: TensorRT\onnx_parser\ModelImporter.cpp:534: Found unsupported shape-tensor producing layer:Floor_1101
[2023-05-15 15:36:40][info][trt_builder.cpp:558]:Input shape is -1 x 256 x 64 x 64
[2023-05-15 15:36:40][info][trt_builder.cpp:559]:Set max batch size = 1
[2023-05-15 15:36:40][info][trt_builder.cpp:560]:Set max workspace size = 3072.00 MB
[2023-05-15 15:36:40][info][trt_builder.cpp:561]:Base device: [ID 0]<NVIDIA GeForce RTX 3060>[arch 8.6][GMEM 10.75 GB/12.00 GB]
[2023-05-15 15:36:40][info][trt_builder.cpp:564]:Network has 6 inputs:
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 0.[image_embeddings] shape is -1 x 256 x 64 x 64
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 1.[point_coords] shape is -1 x -1 x 2
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 2.[point_labels] shape is -1 x -1
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 3.[mask_input] shape is -1 x 1 x 256 x 256
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 4.[has_mask_input] shape is -1
[2023-05-15 15:36:40][info][trt_builder.cpp:570]: 5.[orig_im_size] shape is -1
[2023-05-15 15:36:40][info][trt_builder.cpp:576]:Network has 3 outputs:
[2023-05-15 15:36:41][error][trt_builder.cpp:30]:NVInfer: 2: [shapeContext.cpp::nvinfer1::builder::ShapeContext::op::1661] Error Code 2: Internal Error (Assertion y->type() == signature.operandType(1) failed. )
"""

Environment

TensorRT Version:
8.5.2.2

NVIDIA GPU:
RTX 3060

NVIDIA Driver Version:
Driver Version: 517.02

CUDA Version:
CUDA Version: 11.7

CUDNN Version:
8.6

Have you tried the latest release?:
no

Can this model run on other frameworks? For example run ONNX model with ONNXRuntime (polygraphy run <model.onnx> --onnxrt):
yes, run with onnxruntime is perfect.

### Tasks