LibreGLMOCR0.9b

GLM-OCR weights mirrored for LibreYOLO's ocr task: the GLM-OCR recognizer at the repository root and the PP-DocLayoutV3 layout detector under layout/, so one snapshot holds the whole two-stage document parser.

Source

• Root: mirrored from zai-org/GLM-OCR at commit 2e85a62840ccac27daa451df36c736c4636b8628. Released by Z.ai under the MIT License (per the upstream model card).
• layout/: mirrored from PaddlePaddle/PP-DocLayoutV3_safetensors at commit e4f489dc536556fc1be02b973ba06756e0d4a2ac (config.json, model.safetensors, preprocessor_config.json). Copyright (c) PaddlePaddle Authors, Apache License 2.0.

Modifications

No learned parameters were changed. Files are byte-identical to the upstream snapshots. Not mirrored: the upstream evaluation records, PP-DocLayoutV3's Paddle inference.yml, and both upstream model cards, replaced by this one.

Usage

from libreyolo import LibreVLM

model = LibreVLM("glm-ocr")
result = model.predict("page.png")
print(result.ocr.markdown)

Requires pip install 'libreyolo[vlm]' with transformers >= 5.3.

License

GLM-OCR (root): MIT License. PP-DocLayoutV3 (layout/): Apache License 2.0. See LICENSE and NOTICE.