microsoft / onnxruntime

ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
https://onnxruntime.ai
MIT License
14.79k stars 2.94k forks source link

different results with onnxruntime-gpu-1.10 #14587

Open cqray1990 opened 1 year ago

cqray1990 commented 1 year ago

Describe the issue

when i test OCR CRNN model with onnxruntime-gpu-1.10 ,it gets different recognise results, that's amazing, mode weights is not stable

To reproduce

when i test OCR CRNN model with onnxruntime-gpu-1.10 ,it gets different recognise results, that's amazing, mode weights is not stable

Urgency

when i test OCR CRNN model with onnxruntime-gpu-1.10 ,it gets different recognise results, that's amazing, mode weights is not stable

Platform

Windows

OS Version

win 11

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

1.10 and 1.12

ONNX Runtime API

C#

Architecture

X64

Execution Provider

CUDA

Execution Provider Library Version

cuda 11.4 cudnn 8.2.2

Model File

No response

Is this a quantized model?

No

yufenglee commented 1 year ago

Could you please provide the instructions to repro the issue? Do you have same issue on CPU EP or with latest OnnxRuntime (1.13)