metadata
license: apache-2.0
base_model:
- google/madlad400-3b-mt
pipeline_tag: translation
tags:
- text2text-generation
- text-generation-inference
This is same model as https://huggingface.co/ISoloist1/madlad400-3b-mt-onnx but with folder structure optimized for Transformers.js
ONNX version of google/madlad400-3b-mt
Converted and quantized with optimum-cli
- Convert to ONNX:
optimum-cli onnxruntime export --model google/madlad400-3b-mt <output_path> --legacy
- Quantization:
optimum-cli onnxruntime quantize --onnx_model <input_model_path> -o <output_model_path> --avx512_vnni