GLM-OCR ONNX Models

ONNX exports of zai-org/GLM-OCR for use with ORT in the papers-extract pipeline.

Attribution

The original GLM-OCR model was created by zai-org and is released under the MIT License.

These ONNX files are derived from the original weights with attention fusion optimizations (GQA/MHA surgery) for efficient CUDA inference. See the original model card at zai-org/GLM-OCR for full details.

License

MIT License - see LICENSE for details.

Downloads last month
38
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support