Troubleshoot NVIDIA NIM for Image OCR (NeMo Retriever OCR)#
Use this documentation to troubleshoot issues that arise when you use NVIDIA NIM for Image OCR (NeMo Retriever OCR).
NIM fails to start#
In some cases, a NIM fails to start when you run the docker run command.
Some NIM models require that you accept the license terms on NGC before you can pull the container image and model assets.
To resolve this issue, browse to the model page on the NGC Catalog, read and then click Accept Terms.
After you accept the terms, rerun the docker run command.
For details, refer to Get Started.
OCR accuracy is lower than expected#
On NVIDIA A10G and NVIDIA L4 GPUs, the multilingual OCR v2 TensorRT profiles can produce lower OCR accuracy than the PyTorch profile. This issue is limited to the A10G and L4 multilingual TensorRT profiles. To resolve this issue, configure the NIM to use a PyTorch model profile. For details, refer to Use PyTorch on A10G and L4 GPUs.