NVIDIA Developer Forums

TensorRT with fp16 return nan for all outputs

AI & Data Science Deep Learning (Training & Inference) TensorRT

spolisetty February 3, 2021, 8:53am 3

Hi @keivan.moazami,

Seems no input is fed to TRT, the output could be any value (including NaN).

Thank you.

All outputs are nan

Topic		Replies	Views	Activity
All outputs are nan TensorRT	5	2702	September 23, 2022
Inference fp16 engine in c++ get Nan output but inference fp32 engine can get correct result TensorRT	13	1351	October 10, 2023
ONNX Model and Tensorrt Engine gives different output for parseq model TensorRT onnx	4	1243	July 17, 2023
FP16 gets wrong results with dynamic input TensorRT	3	724	January 13, 2022
Inference TensorRT randomly returns nan TensorRT tensorrt	2	551	April 27, 2023
ONNX to TensorRT conversion (FP16 or FP32) results in integer outputs being mapped to near negative infinity (~2e-45) TensorRT tensorrt , cuda , onnx , aws , natural-language-processing-nlp , nlp	3	3397	June 6, 2022
Use trtexec to convert onnx format to fp16 tensorRT engine, and perform inference and export nan TensorRT cudnn	2	38	June 3, 2025
Wrong result in TensorRT, but it seems something is working correctly TensorRT tensorrt , onnx	9	2455	September 29, 2022
Convert the TRT model with FP16 Jetson TX2 jetpack , tensorrt , jetson-inference	7	2489	October 18, 2021
TensorRT int8 slower than FP16 due to reformat layer TensorRT tensorrt , cudnn	0	97	October 11, 2024