Format conversion onnx, rknn

#12
by litterbearjiujiu - opened

讨论能否转换格式到rk3588,rk3568等嵌入式板端npu上运行加快推理速度

有大佬尝试转换并运行成功吗

Hi @litterbearjiujiu ,

Welcome to Gemma models, you could able to find the ONNX model under the following URL from ONNX community. Please let me know if any additional assistance is required.

To know more about google/embeddinggemma-300m please visit the following ref.

Thanks.

Hi @BalakrishnaCh ! Is there a onnx conversion script ? Can I see it, please ? I'm trying to convert a fine-tuned version and none of these methods work for me: sentence-transformers (pip latest), torch.onnx.export (2.9), optimum-onnx (git), transformers (git, embeddinggemma branch). I would really appreciate some help. Thanks in advance!

Sign up or log in to comment