Release Notes for NVIDIA NIM for Image OCR (NeMo Retriever OCR v1)#
This documentation contains the release notes for NVIDIA NIM for Image OCR (NeMo Retriever OCR v1).
Release 1.2.0#
Summary#
This is the first General Access release of the NVIDIA NIM for Image OCR (NeMo Retriever OCR v1).
Added TRT optimized engines for CUDA GPU Compute Capability. Support includes 12.0, 10.0, 9.0, 8.9, 8.6, and 8.0.
The
NIM_TRITON_OPTIMIZATION_MODEenvironment variable is no longer supported.
Known Issues#
The
persistence.enabledvalue and all related dependent configuration flags are currently non-functional in the NIM helm chart.
Release 1.1.0#
Summary#
Upgraded to use Triton Inference Server 25.08 to address CVEs.
Added Triton Ensemble Configuration which supports configuring the underlying Triton Ensemble model pipeline.
Added the
NIM_TRITON_PINNED_MEMORY_POOL_MBenvironment variable.Added the
NIM_TRITON_ENABLE_MODEL_CONTROLenvironment variable.Added the
NIM_TRITON_IDLE_BYTES_LIMITenvironment variable.Added the
NIM_TRITON_FLUSH_INTERVALenvironment variable.Added the
NIM_TRITON_RATE_LIMITenvironment variable.
Known Issues#
The
persistence.enabledvalue and all related dependent configuration flags are currently non-functional in the NIM helm chart.
Release 1.0.0#
Summary#
This is the first Early Access release of the NVIDIA NIM for Image OCR (NeMo Retriever OCR v1).
Known Issues#
This release only supports a single GPU-agnostic PyTorch backend profile.