BAAI

/

bge-m3

like 866

1. I have already converted the model to ONNX at https://huggingface.co/Xenova/bge-m3
2. The conversion pipeline is available at https://github.com/xenova/transformers.js/blob/5ac17bda838547b9167a75ba1eb1b2e98b680cab/scripts/convert.py

Hope that helps!

Thanks that helps! I'll verify export accuracy and submit it here + instructions

@Xenova what about the fp16 version? How was it created?

@Nemphys I followed this tutorial: https://onnxruntime.ai/docs/performance/model-optimizations/float16.html :)