Accuracy problem between onnx and fp16 trt inference

Description

I am encountering an accuracy discrepancy between ONNX inference and TensorRT FP32 inference.

Environment

TensorRT Version: 10.8.0.43

NVIDIA GPU: RTX 3060

NVIDIA Driver Version: 560.35.05

CUDA Version: 12.4.99

CUDNN Version: 9.8.0.87

Operating System: 24.04.1-Ubuntu

Python Version (if applicable): 3.11.11

PyTorch Version (if applicable): 2.6.0

Relevant Files

Model link:

https://drive.google.com/drive/folders/1OaczMFXSv2a46QZHwSsn6lT8V3S_Ki-O?usp=sharing

Steps To Reproduce

polygraphy run --onnxrt tlr_202506151003.onnx \
    --data-loader-script tool/data_loader.py \
    --save-outputs outputs_fp32.json

trtexec --onnx=tlr_202506151003.onnx --saveEngine=model_fp32.plan

polygraphy run --trt model_fp32.plan \
    --data-loader-script tool/data_loader.py \
    --load-outputs outputs_fp32.json \
    --atol 0.01 --rtol 0.01

Have you tried the latest release?: not yet

Start with tool/data_loader.py and reproduce the ONNX Runtime and TensorRT commands in the issue using the linked model and listed environment. Compare the saved outputs and document the exact mismatch, including whether it occurs with FP32 or FP16; the issue is ready for follow-up when the discrepancy is reproducible and its scope is clear.