TensorRT build engine failed nanoSAM

Hi,

Im trying to build nanoSAM model,
when I build the engine I get the following error:
[E] Error[4]: [graph.cpp::symbolicExecute::539] Error Code 4: Internal Error (/OneHot: an IIOneHotLayer cannot be used to compute a shape tensor).

I tried to convert mobile_sam _decoder model to INT32, Still get the same problem.

The full output:
&&&& RUNNING TensorRT.trtexec [TensorRT v8601] # /usr/src/tensorrt/bin/trtexec --onnx=data/sam32.onnx --saveEngine=data/mobile_sam_mask_decoder.engine --minShapes=point_coords:1x1x2,point_labels:1x1 --optShapes=point_coords:1x1x2,point_labels:1x1 --maxShapes=point_coords:1x10x2,point_labels:1x10
[02/07/2024-08:01:59] [I] === Model Options ===
[02/07/2024-08:01:59] [I] Format: ONNX
[02/07/2024-08:01:59] [I] Model: data/sam32.onnx
[02/07/2024-08:01:59] [I] Output:
[02/07/2024-08:01:59] [I] === Build Options ===
[02/07/2024-08:01:59] [I] Max batch: explicit batch
[02/07/2024-08:01:59] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default
[02/07/2024-08:01:59] [I] minTiming: 1
[02/07/2024-08:01:59] [I] avgTiming: 8
[02/07/2024-08:01:59] [I] Precision: FP32
[02/07/2024-08:01:59] [I] LayerPrecisions:
[02/07/2024-08:01:59] [I] Layer Device Types:
[02/07/2024-08:01:59] [I] Calibration:
[02/07/2024-08:01:59] [I] Refit: Disabled
[02/07/2024-08:01:59] [I] Version Compatible: Disabled
[02/07/2024-08:01:59] [I] TensorRT runtime: full
[02/07/2024-08:01:59] [I] Lean DLL Path:
[02/07/2024-08:01:59] [I] Tempfile Controls: { in_memory: allow, temporary: allow }
[02/07/2024-08:01:59] [I] Exclude Lean Runtime: Disabled
[02/07/2024-08:01:59] [I] Sparsity: Disabled
[02/07/2024-08:01:59] [I] Safe mode: Disabled
[02/07/2024-08:01:59] [I] Build DLA standalone loadable: Disabled
[02/07/2024-08:01:59] [I] Allow GPU fallback for DLA: Disabled
[02/07/2024-08:01:59] [I] DirectIO mode: Disabled
[02/07/2024-08:01:59] [I] Restricted mode: Disabled
[02/07/2024-08:01:59] [I] Skip inference: Disabled
[02/07/2024-08:01:59] [I] Save engine: data/mobile_sam_mask_decoder.engine
[02/07/2024-08:01:59] [I] Load engine:
[02/07/2024-08:01:59] [I] Profiling verbosity: 0
[02/07/2024-08:01:59] [I] Tactic sources: Using default tactic sources
[02/07/2024-08:01:59] [I] timingCacheMode: local
[02/07/2024-08:01:59] [I] timingCacheFile:
[02/07/2024-08:01:59] [I] Heuristic: Disabled
[02/07/2024-08:01:59] [I] Preview Features: Use default preview flags.
[02/07/2024-08:01:59] [I] MaxAuxStreams: -1
[02/07/2024-08:01:59] [I] BuilderOptimizationLevel: -1
[02/07/2024-08:01:59] [I] Input(s)s format: fp32:CHW
[02/07/2024-08:01:59] [I] Output(s)s format: fp32:CHW
[02/07/2024-08:01:59] [I] Input build shape: point_coords=1x1x2+1x1x2+1x10x2
[02/07/2024-08:01:59] [I] Input build shape: point_labels=1x1+1x1+1x10
[02/07/2024-08:01:59] [I] Input calibration shapes: model
[02/07/2024-08:01:59] [I] === System Options ===
[02/07/2024-08:01:59] [I] Device: 0
[02/07/2024-08:01:59] [I] DLACore:
[02/07/2024-08:01:59] [I] Plugins:
[02/07/2024-08:01:59] [I] setPluginsToSerialize:
[02/07/2024-08:01:59] [I] dynamicPlugins:
[02/07/2024-08:01:59] [I] ignoreParsedPluginLibs: 0
[02/07/2024-08:01:59] [I]
[02/07/2024-08:01:59] [I] === Inference Options ===
[02/07/2024-08:01:59] [I] Batch: Explicit
[02/07/2024-08:01:59] [I] Input inference shape: point_labels=1x1
[02/07/2024-08:01:59] [I] Input inference shape: point_coords=1x1x2
[02/07/2024-08:01:59] [I] Iterations: 10
[02/07/2024-08:01:59] [I] Duration: 3s (+ 200ms warm up)
[02/07/2024-08:01:59] [I] Sleep time: 0ms
[02/07/2024-08:01:59] [I] Idle time: 0ms
[02/07/2024-08:01:59] [I] Inference Streams: 1
[02/07/2024-08:01:59] [I] ExposeDMA: Disabled
[02/07/2024-08:01:59] [I] Data transfers: Enabled
[02/07/2024-08:01:59] [I] Spin-wait: Disabled
[02/07/2024-08:01:59] [I] Multithreading: Disabled
[02/07/2024-08:01:59] [I] CUDA Graph: Disabled
[02/07/2024-08:01:59] [I] Separate profiling: Disabled
[02/07/2024-08:01:59] [I] Time Deserialize: Disabled
[02/07/2024-08:01:59] [I] Time Refit: Disabled
[02/07/2024-08:01:59] [I] NVTX verbosity: 0
[02/07/2024-08:01:59] [I] Persistent Cache Ratio: 0
[02/07/2024-08:01:59] [I] Inputs:
[02/07/2024-08:01:59] [I] === Reporting Options ===
[02/07/2024-08:01:59] [I] Verbose: Disabled
[02/07/2024-08:01:59] [I] Averages: 10 inferences
[02/07/2024-08:01:59] [I] Percentiles: 90,95,99
[02/07/2024-08:01:59] [I] Dump refittable layers:Disabled
[02/07/2024-08:01:59] [I] Dump output: Disabled
[02/07/2024-08:01:59] [I] Profile: Disabled
[02/07/2024-08:01:59] [I] Export timing to JSON file:
[02/07/2024-08:01:59] [I] Export output to JSON file:
[02/07/2024-08:01:59] [I] Export profile to JSON file:
[02/07/2024-08:01:59] [I]
[02/07/2024-08:01:59] [I] === Device Information ===
[02/07/2024-08:01:59] [I] Selected Device: Quadro M4000
[02/07/2024-08:01:59] [I] Compute Capability: 5.2
[02/07/2024-08:01:59] [I] SMs: 13
[02/07/2024-08:01:59] [I] Device Global Memory: 8112 MiB
[02/07/2024-08:01:59] [I] Shared Memory per SM: 96 KiB
[02/07/2024-08:01:59] [I] Memory Bus Width: 256 bits (ECC disabled)
[02/07/2024-08:01:59] [I] Application Compute Clock Rate: 0.7725 GHz
[02/07/2024-08:01:59] [I] Application Memory Clock Rate: 3.005 GHz
[02/07/2024-08:01:59] [I]
[02/07/2024-08:01:59] [I] Note: The application clock rates do not reflect the actual clock rates that the GPU is currently running at.
[02/07/2024-08:01:59] [I]
[02/07/2024-08:01:59] [I] TensorRT version: 8.6.1
[02/07/2024-08:01:59] [I] Loading standard plugins
[02/07/2024-08:01:59] [I] [TRT] [MemUsageChange] Init CUDA: CPU +2, GPU +0, now: CPU 13, GPU 681 (MiB)
[02/07/2024-08:02:09] [I] [TRT] [MemUsageChange] Init builder kernel library: CPU +128, GPU +0, now: CPU 217, GPU 681 (MiB)
[02/07/2024-08:02:09] [I] Start parsing network model.
[02/07/2024-08:02:09] [I] [TRT] ----------------------------------------------------------------
[02/07/2024-08:02:09] [I] [TRT] Input filename: data/sam32.onnx
[02/07/2024-08:02:09] [I] [TRT] ONNX IR version: 0.0.9
[02/07/2024-08:02:09] [I] [TRT] Opset version: 16
[02/07/2024-08:02:09] [I] [TRT] Producer name: onnx-typecast
[02/07/2024-08:02:09] [I] [TRT] Producer version:
[02/07/2024-08:02:09] [I] [TRT] Domain:
[02/07/2024-08:02:09] [I] [TRT] Model version: 0
[02/07/2024-08:02:09] [I] [TRT] Doc string:
[02/07/2024-08:02:09] [I] [TRT] ----------------------------------------------------------------
[02/07/2024-08:02:09] [W] [TRT] onnx2trt_utils.cpp:374: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[02/07/2024-08:02:09] [E] Error[4]: [graph.cpp::symbolicExecute::539] Error Code 4: Internal Error (/OneHot: an IIOneHotLayer cannot be used to compute a shape tensor)
[02/07/2024-08:02:09] [E] [TRT] ModelImporter.cpp:771: While parsing node number 146 [Tile → “/Tile_output_0”]:
[02/07/2024-08:02:09] [E] [TRT] ModelImporter.cpp:772: — Begin node —
[02/07/2024-08:02:09] [E] [TRT] ModelImporter.cpp:773: input: “/Unsqueeze_3_output_0”
input: “/Reshape_2_output_0”
output: “/Tile_output_0”
name: “/Tile”
op_type: “Tile”

[02/07/2024-08:02:09] [E] [TRT] ModelImporter.cpp:774: — End node —
[02/07/2024-08:02:09] [E] [TRT] ModelImporter.cpp:777: ERROR: ModelImporter.cpp:195 In function parseGraph:
[6] Invalid Node - /Tile
[graph.cpp::symbolicExecute::539] Error Code 4: Internal Error (/OneHot: an IIOneHotLayer cannot be used to compute a shape tensor)
[02/07/2024-08:02:09] [E] Failed to parse onnx file
[02/07/2024-08:02:09] [I] Finished parsing network model. Parse time: 0.0837755
[02/07/2024-08:02:09] [E] Parsing model failed
[02/07/2024-08:02:09] [E] Failed to create engine from model or file.
[02/07/2024-08:02:09] [E] Engine set up failed
&&&& FAILED TensorRT.trtexec [TensorRT v8601] # /usr/src/tensorrt/bin/trtexec --onnx=data/sam32.onnx --saveEngine=data/mobile_sam_mask_decoder.engine --minShapes=point_coords:1x1x2,point_labels:1x1 --optShapes=point_coords:1x1x2,point_labels:1x1 --maxShapes=point_coords:1x10x2,point_labels:1x10