Enable PDF extraction with Nemoretriever Parse for NVIDIA RAG Blueprint#

For enhanced PDF extraction capabilities, you can use the Nemoretriever Parse service with the NVIDIA RAG Blueprint. This service provides improved PDF parsing and structure understanding compared to the default PDF extraction method.

Warning

B200 GPUs are not supported for PDF extraction with Nemoretriever Parse. For this feature, use H100 or A100 GPUs instead.

Using Docker Compose#

Using On-Prem Models#

  1. Prerequisites: Follow the deployment guide up to and including the step labelled “Start all required NIMs.”

  2. Deploy the Nemoretriever Parse service along with other required NIMs:

    USERID=$(id -u) docker compose --profile rag --profile nemoretriever-parse -f deploy/compose/nims.yaml up -d

  3. Configure the ingestor-server to use Nemoretriever Parse by setting the environment variable:

    export APP_NVINGEST_PDFEXTRACTMETHOD=nemoretriever_parse

  4. Deploy the ingestion-server and rag-server containers following the remaining steps in the deployment guide.

  5. You can now ingest PDF files using the ingestion API usage notebook.

Using NVIDIA Hosted API Endpoints#

  1. Prerequisites: Follow the deployment guide up to and including the step labelled “Start the vector db containers from the repo root.”

  2. Export the following variables to use nemoretriever parse API endpoints:

    export NEMORETRIEVER_PARSE_HTTP_ENDPOINT=https://integrate.api.nvidia.com/v1/chat/completions
export NEMORETRIEVER_PARSE_MODEL_NAME=nvidia/nemoretriever-parse
export NEMORETRIEVER_PARSE_INFER_PROTOCOL=http

  3. Configure the ingestor-server to use Nemoretriever Parse by setting the environment variable:

    export APP_NVINGEST_PDFEXTRACTMETHOD=nemoretriever_parse

  4. Deploy the ingestion-server and rag-server containers following the remaining steps in the deployment guide.

  5. You can now ingest PDF files using the ingestion API usage notebook.

Note

When using NVIDIA hosted endpoints, you may encounter rate limiting with larger file ingestions (>10 files).

Using Helm#

To enable PDF extraction with Nemoretriever Parse using Helm, you need to enable the Nemoretriever Parse service along with other required services:

helm upgrade --install rag -n rag https://helm.ngc.nvidia.com/0648981100760671/charts/nvidia-blueprint-rag-v2.4.0-dev.tgz \
  --username '$oauthtoken' \
  --password "${NGC_API_KEY}" \
  --set imagePullSecret.password=$NGC_API_KEY \
  --set ngcApiSecret.password=$NGC_API_KEY \
  --set nv-ingest.nim-vlm-text-extraction.deployed=true \
  --set ingestor-server.envVars.APP_NVINGEST_PDFEXTRACTMETHOD="nemoretriever_parse"

Limitations and Requirements#

When using Nemoretriever Parse for PDF extraction, consider the following:

  • Nemoretriever Parse only supports PDF format documents. Attempting to process non-PDF files will lead them to be extracted using the default extraction method.

  • The service requires GPU resources. Make sure you have sufficient GPU resources available before enabling this feature.

  • The extraction quality may vary depending on the PDF structure and content.

  • Nemoretriever Parse is currently not supported on NVIDIA B200 GPUs.

For detailed information about hardware requirements and supported GPUs for all NeMo Retriever extraction NIMs, refer to the NeMo Retriever Extraction Support Matrix.

Available PDF Extraction Methods#

The APP_NVINGEST_PDFEXTRACTMETHOD environment variable supports the following values:

  • nemoretriever_parse: Uses the Nemoretriever Parse service for enhanced PDF extraction

  • pdfium: Uses the default PDFium-based extraction

  • None: Uses the default extraction method

Note

The Nemoretriever Parse service requires GPU resources. Make sure you have sufficient GPU resources available before enabling this feature.