DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines

• URL: http://arxiv.org/abs/2604.14314v1
• Date: Wed, 15 Apr 2026 18:17:11 GMT
• Title: DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines
• Abstract summary: This manuscript introduces DharmaOCR Full and Lite, a pair of specialized small language models (SSLMs) for structured OCR.<n>It also presents DharmaOCR-Benchmark, a benchmark that covers printed, handwritten, and legal/administrative documents.<n>It proposes a unified evaluation protocol that measures fidelity and structure while explicitly tracking text degeneration.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: This manuscript introduces DharmaOCR Full and Lite, a pair of specialized small language models (SSLMs) for structured OCR that jointly optimize transcription quality, generation stability, and inference cost. It also presents DharmaOCR-Benchmark, a benchmark that covers printed, handwritten, and legal/administrative documents, and proposes a unified evaluation protocol that measures fidelity and structure while explicitly tracking text degeneration as a first-class benchmark metric (alongside unit cost). Beyond reporting degeneration rates, the manuscript empirically shows degeneration is not merely a quality failure, since it materially worsens production performance by increasing response time, reducing throughput, and inflating computational cost due to abnormally long generations. To the best of the author's knowledge, as a methodological contribution, this is the first application of Direct Preference Optimization (DPO) for OCR, explicitly using degenerate generations as rejected examples to penalize looping behavior. Combined with Supervised Fine-Tuning (SFT) for enforcing a strict JSON schema (header, margin, footer, and text), DPO consistently reduces degeneration rate across model families (up to 87.6% relative) while preserving or improving extraction quality. The resulting models, namely, DharmaOCR Full (7B) and DharmaOCR Lite (3B), set a new state-of-the-art on DharmaOCR-Benchmark, outperforming each open-source and commercial baseline model evaluated regarding extraction quality, reaching 0.925 and 0.911 scores with 0.40% and 0.20% degeneration rates. AWQ quantization reduced up to 22% per-page cost with negligible quality loss, enabling a strong quality-cost trade-off in comparison to proprietary OCR APIs and open-source alternatives.

Related papers

• When Good OCR Is Not Enough: Benchmarking OCR Robustness for Retrieval-Augmented Generation [2.8437065210971872]
  Industrial Retrieval-Augmented Generation (RAG) systems depend on optical character recognition (OCR) to transform visual documents into text.<n>Existing OCR benchmarks rely on character-level metrics, which inadequately measure downstream RAG effectiveness under real-world conditions.<n>We introduce an OCR benchmark for industrial RAG systems covering 11 challenging document types.
  arXiv  Detail & Related papers  (2026-04-29T09:42:42Z)
• GLM-OCR Technical Report [65.42028025507491]
  GLM-OCR is an efficient compact model designed for real-world document understanding.<n>It combines a CogViT visual encoder with a GLM language decoder, achieving a strong balance between computational efficiency and recognition performance.<n>Extensive evaluations on public benchmarks and industrial scenarios show that GLM-OCR achieves competitive or state-of-the-art performance.
  arXiv  Detail & Related papers  (2026-03-11T15:55:47Z)
• CDRRM: Contrast-Driven Rubric Generation for Reliable and Interpretable Reward Modeling [61.75914342638658]
  CDRRM (Contrast-Driven Reward Model) is a framework for high-quality rubric generation and guided preference judgment.<n>This work offers a scalable, interpretable, and data-efficient path for reward modeling.
  arXiv  Detail & Related papers  (2026-03-09T07:15:23Z)
• Reward-Guided Speculative Decoding for Efficient LLM Reasoning [80.55186052123196]
  We introduce Reward-Guided Speculative Decoding (RSD), a novel framework aimed at improving the efficiency of inference in large language models (LLMs)<n>RSD incorporates a controlled bias to prioritize high-reward outputs, in contrast to existing speculative decoding methods that enforce strict unbiasedness.<n>RSD delivers significant efficiency gains against decoding with the target model only, while achieving significant better accuracy than parallel decoding method on average.
  arXiv  Detail & Related papers  (2025-01-31T17:19:57Z)
• Bridging SFT and DPO for Diffusion Model Alignment with Self-Sampling Preference Optimization [67.8738082040299]
  Self-Sampling Preference Optimization (SSPO) is a new alignment method for post-training reinforcement learning.<n>SSPO eliminates the need for paired data and reward models while retaining the training stability of SFT.<n>SSPO surpasses all previous approaches on the text-to-image benchmarks and demonstrates outstanding performance on the text-to-video benchmarks.
  arXiv  Detail & Related papers  (2024-10-07T17:56:53Z)
• CLOCR-C: Context Leveraging OCR Correction with Pre-trained Language Models [0.0]
  This paper introduces Context Leveraging OCR Correction (CLOCR-C)<n>It uses the infilling and context-adaptive abilities of transformer-based language models (LMs) to improve OCR quality.<n>The study aims to determine if LMs can perform post-OCR correction, improve downstream NLP tasks, and the value of providing the socio-cultural context as part of the correction process.
  arXiv  Detail & Related papers  (2024-08-30T17:26:05Z)
• Democratizing LLMs: An Exploration of Cost-Performance Trade-offs in Self-Refined Open-Source Models [53.859446823312126]
  SoTA open source models of varying sizes from 7B - 65B, on average, improve 8.2% from their baseline performance. Strikingly, even models with extremely small memory footprints, such as Vicuna-7B, show a 11.74% improvement overall and up to a 25.39% improvement in high-creativity, open ended tasks.
  arXiv  Detail & Related papers  (2023-10-11T15:56:00Z)
• Quality-Aware Translation Models: Efficient Generation and Quality Estimation in a Single Model [77.19693792957614]
  We propose to make neural machine translation (NMT) models quality-aware by training them to estimate the quality of their own output. We obtain quality gains similar or even superior to quality reranking approaches, but with the efficiency of single pass decoding.
  arXiv  Detail & Related papers  (2023-10-10T15:33:51Z)
• Donut: Document Understanding Transformer without OCR [17.397447819420695]
  We propose a novel VDU model that is end-to-end trainable without underpinning OCR framework. Our approach achieves state-of-the-art performance on various document understanding tasks in public benchmark datasets and private industrial service datasets.
  arXiv  Detail & Related papers  (2021-11-30T18:55:19Z)
• Rerunning OCR: A Machine Learning Approach to Quality Assessment and Enhancement Prediction [0.0]
  Iterating with new and improved OCR solutions enforces decisions to be taken when it comes to targeting the right reprocessing candidates. This article captures the efforts of the National Library of Luxembourg to support those exact decisions.
  arXiv  Detail & Related papers  (2021-10-04T18:52:59Z)
• Lights, Camera, Action! A Framework to Improve NLP Accuracy over OCR documents [2.6201102730518606]
  We demonstrate an effective framework for mitigating OCR errors for any downstream NLP task. We first address the data scarcity problem for model training by constructing a document synthesis pipeline. For the benefit of the community, we have made the document synthesis pipeline available as an open-source project.
  arXiv  Detail & Related papers  (2021-08-06T00:32:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.