Function: createPrivacyFilter()

createPrivacyFilter<Label>(config, runtime?): Promise<PrivacyFilter<Label>>

Defined in: extensions/nlp/tasks/privacyFilter.ts:129

Creates a privacy filter runner that detects personally identifiable information (PII) spans in text.

It loads the tokenizer and model, validates the forward signature against the configured label space, pre-computes the BIOES grammar tables, and registers clean disposal hooks to clear all native memory.

Works with any privacy-filter-style model exporting forward(input_ids, attention_mask) -> logits over a BIOES label space. Inputs longer than the model's exported window are processed in sliding windows with 50% overlap and never truncated; predictions near a window's edges are discarded in favor of the neighboring window's more centered context.

Type Parameters

Label

Label extends string

The model's BIOES label space.

Parameters

config

PrivacyFilterModel<Label>

Privacy filter task configuration containing the model and tokenizer paths plus the label space options. See PrivacyFilterModel.

runtime?

WorkletRuntime

Optional worklet runtime thread on which to run the model execution.

Returns

Promise<PrivacyFilter<Label>>

A promise resolving to the instantiated PrivacyFilter runner.

Throws

With code INVALID_ARGUMENT if labelNames is empty or does not start with 'O', LOAD_FAILED if the model or tokenizer fails to load, or SCHEMA_MISMATCH if the loaded model schema does not match the privacy filter specification.