Dharma AI releases Dharma Field Extractor for document-to-JSON extraction
Overview
Dharma AI releases Dharma Field Extractor, a 2-billion-parameter model that converts document images into structured JSON fields.
Users can either request all fields without a schema or supply an exact schema, and missing values are returned as null. The model is a LoRA adapter on DharmaOCR-LITE v1.1, trained on 43,384 examples.
On a 546-document benchmark, Dharma says the model outperforms four frontier LLM configurations on field-level F1 at lower cost. The benchmark results are Dharma's own claims and have not been independently verified.
Written by AI from the articles below · updated Nov 4, 1:49 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Hugging Face BlogOfficialDharma AI releases Field Extractor, a 2B model that outputs structured JSON from documents
AIDharma AI releases Dharma Field Extractor, a 2-billion-parameter model that turns any document image into structured JSON fields. Users can request all fields with no schema or supply an exact schema, with null returned for missing values. The model is a LoRA adapter on DharmaOCR-LITE v1.1, trained on 43,384 examples; on a 546-document benchmark, Dharma says it outperforms four frontier LLM configurations on field-level F1 at lower cost.
Heat Trend
Not enough continuous observations to show a trend yet.