Fail to make TensorRT model from onnx in Isaac ROS Image Segment Tutorial

https://nvidia-isaac-ros.github.io/repositories_and_packages/isaac_ros_image_segmentation/isaac_ros_segformer/index.html#quickstart
I fellow the tutorial and the command "/usr/src/tensorrt/bin/trtexec --onnx=${ISAAC_ROS_WS}/isaac_ros_assets/models/peoplesemsegformer/1/model.onnx --saveEngine=${ISAAC_ROS_WS}/isaac_ros_assets/models/peoplesemsegformer/1/model.plan"

show the error:

(isaac-ros) jim@jim-ROG-Strix-G834JZ-G834JZ:~/workspaces/isaac_ros-dev/isaac_ros_assets/models/peoplesemsegformer/1$ cd ${ISAAC_ROS_WS}/isaac_ros_assets/models/peoplesemsegformer/1

rm -f model_int64.plan

/usr/src/tensorrt/bin/trtexec
--onnx=model.onnx
--saveEngine=model_int64.plan
--shapes=input:1x3x512x512
--outputIOFormats=int64:chw
2>&1 | tee /tmp/trtexec_int64.log

ls -lh model_int64.plan
&&&& RUNNING TensorRT.trtexec [TensorRT v101601] [b11] # /usr/src/tensorrt/bin/trtexec --onnx=model.onnx --saveEngine=model_int64.plan --shapes=input:1x3x512x512 --outputIOFormats=int64:chw
[05/04/2026-16:47:41] [I] TF32 is enabled by default. Add --noTF32 flag to further improve accuracy with some performance cost.
[05/04/2026-16:47:41] [I] === Model Options ===
[05/04/2026-16:47:41] [I] Format: ONNX
[05/04/2026-16:47:41] [I] Model: model.onnx
[05/04/2026-16:47:41] [I] Output:
[05/04/2026-16:47:41] [I] === Build Options ===
[05/04/2026-16:47:41] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default, tacticSharedMem: default
[05/04/2026-16:47:41] [I] avgTiming: 8
[05/04/2026-16:47:41] [I] Precision: FP32
[05/04/2026-16:47:41] [I] LayerPrecisions:
[05/04/2026-16:47:41] [I] Layer Device Types:
[05/04/2026-16:47:41] [I] Decomposable Attentions:
[05/04/2026-16:47:41] [I] Calibration:
[05/04/2026-16:47:41] [I] Refit: Disabled
[05/04/2026-16:47:41] [I] Strip weights: Disabled
[05/04/2026-16:47:41] [I] Version Compatible: Disabled
[05/04/2026-16:47:41] [I] ONNX Plugin InstanceNorm: Disabled
[05/04/2026-16:47:41] [I] ONNX kENABLE_UINT8_AND_ASYMMETRIC_QUANTIZATION_DLA flag: Disabled
[05/04/2026-16:47:41] [I] ONNX kREPORT_CAPABILITY_DLA flag: Disabled
[05/04/2026-16:47:41] [I] ONNX kADJUST_FOR_DLA flag: Disabled
[05/04/2026-16:47:41] [I] ONNX kENABLE_PLUGIN_OVERRIDE flag: Disabled
[05/04/2026-16:47:41] [I] TensorRT runtime: full
[05/04/2026-16:47:41] [I] Lean DLL Path:
[05/04/2026-16:47:41] [I] Tempfile Controls: { in_memory: allow, temporary: allow }
[05/04/2026-16:47:41] [I] Exclude Lean Runtime: Disabled
[05/04/2026-16:47:41] [I] Sparsity: Disabled
[05/04/2026-16:47:41] [I] Safe mode: Disabled
[05/04/2026-16:47:41] [I] Build DLA standalone loadable: Disabled
[05/04/2026-16:47:41] [I] Allow GPU fallback for DLA: Disabled
[05/04/2026-16:47:41] [I] DirectIO mode: Disabled
[05/04/2026-16:47:41] [I] Restricted mode: Disabled
[05/04/2026-16:47:41] [I] Skip inference: Disabled
[05/04/2026-16:47:41] [I] Save engine: model_int64.plan
[05/04/2026-16:47:41] [I] Load engine:
[05/04/2026-16:47:41] [I] Profiling verbosity: 0
[05/04/2026-16:47:41] [I] Tactic sources: Using default tactic sources
[05/04/2026-16:47:41] [I] timingCacheMode: local
[05/04/2026-16:47:41] [I] timingCacheFile:
[05/04/2026-16:47:41] [I] Enable Compilation Cache: Enabled
[05/04/2026-16:47:41] [I] Enable Monitor Memory: Disabled
[05/04/2026-16:47:41] [I] CPU Only Mode: Disabled
[05/04/2026-16:47:41] [I] errorOnTimingCacheMiss: Disabled
[05/04/2026-16:47:41] [I] Preview Features: Use default preview flags.
[05/04/2026-16:47:41] [I] MaxAuxStreams: -1
[05/04/2026-16:47:41] [I] BuilderOptimizationLevel: -1
[05/04/2026-16:47:41] [I] MaxTactics: -1
[05/04/2026-16:47:41] [I] Calibration Profile Index: 0
[05/04/2026-16:47:41] [I] Weight Streaming: Disabled
[05/04/2026-16:47:41] [I] Runtime Platform: Same As Build
[05/04/2026-16:47:41] [I] Debug Tensors:
[05/04/2026-16:47:41] [I] Distributive Independence: Disabled
[05/04/2026-16:47:41] [I] Mark Unfused Tensors As Debug Tensors: Disabled
[05/04/2026-16:47:41] [I] Input(s)s format: fp32:CHW
[05/04/2026-16:47:41] [I] Output(s): int64:chw
[05/04/2026-16:47:41] [I] Input build shape (profile 0): input=1x3x512x512+1x3x512x512+1x3x512x512
[05/04/2026-16:47:41] [I] Input calibration shapes: model
[05/04/2026-16:47:41] [I] === System Options ===
[05/04/2026-16:47:41] [I] Device: 0
[05/04/2026-16:47:41] [I] DLACore:
[05/04/2026-16:47:41] [I] Loaded static plugins:
[05/04/2026-16:47:41] [I] setPluginsToSerialize:
[05/04/2026-16:47:41] [I] dynamicPlugins:
[05/04/2026-16:47:41] [I] ignoreParsedPluginLibs: 0
[05/04/2026-16:47:41] [I]
[05/04/2026-16:47:41] [I] === Inference Options ===
[05/04/2026-16:47:41] [I] Batch: Explicit
[05/04/2026-16:47:41] [I] Input inference shape : input=1x3x512x512
[05/04/2026-16:47:41] [I] Iterations: 10
[05/04/2026-16:47:41] [I] Duration: 3s (+ 200ms warm up)
[05/04/2026-16:47:41] [I] Sleep time: 0ms
[05/04/2026-16:47:41] [I] Idle time: 0ms
[05/04/2026-16:47:41] [I] Inference Streams: 1
[05/04/2026-16:47:41] [I] ExposeDMA: Disabled
[05/04/2026-16:47:41] [I] Data transfers: Enabled
[05/04/2026-16:47:41] [I] Spin-wait: Disabled
[05/04/2026-16:47:41] [I] Multithreading: Disabled
[05/04/2026-16:47:41] [I] CUDA Graph: Disabled
[05/04/2026-16:47:41] [I] Separate profiling: Disabled
[05/04/2026-16:47:41] [I] Time Deserialize: Disabled
[05/04/2026-16:47:41] [I] Time Refit: Disabled
[05/04/2026-16:47:41] [I] NVTX verbosity: 0
[05/04/2026-16:47:41] [I] Persistent Cache Ratio: 0
[05/04/2026-16:47:41] [I] Optimization Profile Index: 0
[05/04/2026-16:47:41] [I] Weight Streaming Budget: 100.000000%
[05/04/2026-16:47:41] [I] Inputs:
[05/04/2026-16:47:41] [I] Debug Tensor Save Destinations:
[05/04/2026-16:47:41] [I] Dump All Debug Tensor in Formats:
[05/04/2026-16:47:41] [I] === Reporting Options ===
[05/04/2026-16:47:41] [I] Verbose: Disabled
[05/04/2026-16:47:41] [I] Averages: 10 inferences
[05/04/2026-16:47:41] [I] Percentiles: 90,95,99
[05/04/2026-16:47:41] [I] Dump refittable layers:Disabled
[05/04/2026-16:47:41] [I] Dump output: Disabled
[05/04/2026-16:47:41] [I] Profile: Disabled
[05/04/2026-16:47:41] [I] Export timing to JSON file:
[05/04/2026-16:47:41] [I] Export output to JSON file:
[05/04/2026-16:47:41] [I] Export profile to JSON file:
[05/04/2026-16:47:41] [I]
[05/04/2026-16:47:41] [I] === Device Information ===
[05/04/2026-16:47:41] [I] Available Devices:
[05/04/2026-16:47:41] [I] Device 0: "NVIDIA GeForce RTX 4080 Laptop GPU" UUID: GPU-7d2d621d-c5b1-2cd9-91df-d67292816df5
[05/04/2026-16:47:41] [I] Selected Device: NVIDIA GeForce RTX 4080 Laptop GPU
[05/04/2026-16:47:41] [I] Selected Device ID: 0
[05/04/2026-16:47:41] [I] Selected Device UUID: GPU-7d2d621d-c5b1-2cd9-91df-d67292816df5
[05/04/2026-16:47:41] [I] Compute Capability: 8.9
[05/04/2026-16:47:41] [I] SMs: 58
[05/04/2026-16:47:41] [I] Device Global Memory: 11874 MiB
[05/04/2026-16:47:41] [I] Shared Memory per SM: 100 KiB
[05/04/2026-16:47:41] [I] Memory Bus Width: 192 bits (ECC disabled)
[05/04/2026-16:47:41] [I] Application Compute Clock Rate: 2.28 GHz
[05/04/2026-16:47:41] [I] Application Memory Clock Rate: 9.001 GHz
[05/04/2026-16:47:41] [I]
[05/04/2026-16:47:41] [I] Note: The application clock rates do not reflect the actual clock rates that the GPU is currently running at.
[05/04/2026-16:47:41] [I]
[05/04/2026-16:47:41] [I] TensorRT version: 10.16.1
[05/04/2026-16:47:41] [I] Loading standard plugins
[05/04/2026-16:47:41] [I] [TRT] [MemUsageChange] Init CUDA: CPU +2, GPU +0, now: CPU 32, GPU 249 (MiB)
[05/04/2026-16:47:43] [I] [TRT] [MemUsageChange] Init builder kernel library: CPU +2775, GPU +446, now: CPU 3009, GPU 695 (MiB)
[05/04/2026-16:47:43] [I] Start parsing network model.
[05/04/2026-16:47:43] [I] [TRT] ----------------------------------------------------------------
[05/04/2026-16:47:43] [I] [TRT] Input filename: model.onnx
[05/04/2026-16:47:43] [I] [TRT] ONNX IR version: 0.0.8
[05/04/2026-16:47:43] [I] [TRT] Opset version: 16
[05/04/2026-16:47:43] [I] [TRT] Producer name: pytorch
[05/04/2026-16:47:43] [I] [TRT] Producer version: 1.14.0
[05/04/2026-16:47:43] [I] [TRT] Domain:
[05/04/2026-16:47:43] [I] [TRT] Model version: 0
[05/04/2026-16:47:43] [I] [TRT] Doc string:
[05/04/2026-16:47:43] [I] [TRT] ----------------------------------------------------------------
[05/04/2026-16:47:43] [W] [TRT] ModelImporter.cpp:804: Make sure output output has Int64 binding.
[05/04/2026-16:47:43] [I] Finished parsing network model. Parse time: 0.188357
[05/04/2026-16:47:43] [I] Set shape of input tensor input for optimization profile 0 to: MIN=1x3x512x512 OPT=1x3x512x512 MAX=1x3x512x512
[05/04/2026-16:47:43] [I] Created engine with size: 0 MiB
[05/04/2026-16:47:43] [I] Engine built in 6.722e-05 sec.
[05/04/2026-16:47:43] [E] Assertion failure: false && "Attempting to access an empty engine!"
-rw-rw-r-- 1 jim jim 0 May 4 16:47 model_int64.plan