A new arXiv preprint presents a multi-expert OCR system for historical Manchu manuscripts, routing pages by visual style (regular script, running script, chancery hand) using a lightweight image classifier. The router achieves 99.3% page-level domain accuracy and matches a domain-label oracle, with character error rates of 0.30%, 1.57%, and 4.83% across three script styles. The approach reuses fine-tuning checkpoints as domain specialists, avoiding the need to train dedicated experts for each domain. The work addresses low-resource OCR for endangered historical scripts with a reproducible evaluation protocol.
AlphaOracle is a multi-stage deep learning framework for deciphering oracle bone script (OBS), modeled on expert philological workflows including morphological analysis, contextual retrieval, and classical source validation. Evaluated on 86 domain specialists, it reduced analysis time by 64% and achieved high usefulness ratings, with readings strongly agreeing with expert interpretations. The system also proposes a concrete resolution for the character 'Lao' with implications for Shang dynasty administrative history. The work demonstrates that AI systems aligned with domain expert workflows can advance progress on long-standing undeciphered script problems.
TNG Technology Consulting describes a fine-tuning approach applied to olmOCR, a vision-language model designed for document OCR tasks, to improve its faithfulness and reduce hallucinations. The post covers dataset construction, training methodology, and evaluation results showing improved accuracy on document extraction benchmarks. This represents a practical community contribution to the open-weights document-understanding ecosystem.
GLM-OCR is an open-source OCR project from zai-org built on the GLM model family, positioning itself as accurate, fast, and comprehensive. The repository has accumulated 6,787 GitHub stars with 82 added today, indicating notable community traction. It represents an application of large language/vision models to document understanding and text recognition tasks.
A new arXiv preprint presents a training-free, graph-based method for inferring reading order in complex historical document layouts, including the challenging Glossa Ordinaria manuscript format where text and commentary are spatially interleaved. The approach scores edges in a directed candidate-transition graph using lightweight language model signals (causal LM likelihood and BERT NSP) and recovers global reading order via a degree-constrained directed path cover with a max-regret inference rule. On wrap-around Glossa layouts the method achieves 95% edge accuracy versus 50% for XY-cut, and 88% versus 75% for XY-cut and 25% for LayoutReader on OmniDocBench multi-column pages. The work is relevant to document digitization pipelines and OCR post-processing for historical archives.
Researchers introduce OCR-Robust, a benchmark of 812 samples designed to evaluate how vision-language models handle OCR-reasoning tasks under controlled visual degradation. The benchmark covers documents, scene text, charts, geometry, and tables, applying 5 perturbation types at 3 severity levels each, and evaluates 18 models using metrics including Relative Corruption Retention and a composite Corruption Robustness Index. Key findings show that higher clean accuracy does not guarantee robustness, and that chart and table inputs are substantially more fragile under perturbation than document-like inputs.
PaddlePaddle has released PP-OCRv6 on Hugging Face, an OCR system supporting 50 languages with model sizes ranging from 1.5M to 34.5M parameters. The release spans a wide efficiency-accuracy tradeoff range, making it relevant for both edge and server deployment scenarios. This is a practical open-weights OCR tooling release with multilingual coverage.
A new arXiv preprint introduces Localized LoRA-MoE, a parameter-efficient fine-tuning framework that combines spatial blocking with dynamic context-conditioned routing to address gradient interference in multi-task LoRA adapters. The authors propose two architectures: Block-Wise LoRA-MoE with centralized macro-routing and Cell-Wise LoRA-MoE with decentralized per-cell expert gating. Empirical results across SVD matrix simulations, tabular tasks, and spatial vision benchmarks show that decentralized cell-level gating matches a global coordinator while providing fault isolation against sensor degradation and task-switching instability.
A new arXiv paper introduces a route-specialized dual-adapter architecture for knowledge editing in LLMs, separating the concerns of writing edits (edit adapter) and suppressing them when irrelevant (locality adapter). A relevance router gates which adapter is applied, addressing the locality problem in memory-assisted editing. Evaluated on CounterFact, zsRE, and MQuAKE benchmarks using Llama-3.1-8B-Instruct and Qwen3-8B, the method achieves best-in-class probability-preference accuracy across all three datasets. Ablations show the gain comes from the architectural separation rather than increased parameter capacity.