Generative OCR task

The Generative OCR task uses a vision language model (VLM) to improve the fidelity of text blocks that Unstructured initially processed during partitioning.

Type: prompter

Subtype: see Settings

For more information, see Generative OCR optimization.

Unstructured can produce generative OCR optimizations for workflows that are configured as follows:

• With a Partitioner node set to use the Auto or High Res partitioning strategy, and a generative OCR optimizations node is added.
• With a Partitioner node set to use the VLM partitioning strategy. No generative OCR optimization node is needed (or allowed).

Unstructured never produces generative OCR optimizations for workflows with a Partitioner node set to use the Fast partitioning strategy.

Generative OCR does not process any text blocks by default. You must also explicitly specify which document element types containing text that you want generative OCR to process.

To do this, in a workflow’s High Res partitioner node, add each document element type that you want generative OCR to process to the extract_image_block_types list within the partitioner node’s settings field definition.

Generative OCR does not process the text of any Image or Table elements if they have already been processed by image description or table description enrichments, respectively.

Settings

subtype · string · required

Enrichment algorithm and provider. Set at the WorkflowNode level, outside of settings. Allowed values:

• anthropic_ocr
• bedrock_ocr
• openai_ocr

The preceding list applies only to Unstructured Let’s Go and Pay-As-You-Go accounts.

For Unstructured Business accounts, to get your current list of available values, contact your Unstructured account administrator or Unstructured sales representative, or request support.

provider_type · string · required

Provider that matches the prefix of subtype. Allowed values: anthropic, bedrock, openai.

model · string · required

Model to use for generative OCR. For a full list of the models available in Unstructured, see Available models.

Python SDK

generative_ocr_enrichment_workflow_node = WorkflowNode(
    name="Enrichment",
    subtype="<subtype>",
    type="prompter",
    settings={
        "provider_type": "<provider-type>",
        "model": "<model>"
    }
)

{
    "name": "Enrichment",
    "type": "prompter",
    "subtype": "<subtype>",
    "settings": {
        "provider_type": "<provider-type>",
        "model": "<model>"
    }
}