Anush008 / fastembed-rs

Library for generating vector embeddings, reranking in Rust
https://docs.rs/fastembed
Apache License 2.0
264 stars 36 forks source link

Unable to use ONNX model exported by Optimum #81

Closed lawrenceadams closed 3 months ago

lawrenceadams commented 3 months ago

I wish to use model on huggingface that exist as PyTorch.bin files, however fastembed-rs does not support these if using optimum to convert these.

Optimum tends to export the last layer as logits and not last_hidden_state: but at present fastembed-rs is hardcoded to use the latter.

Desired outcome is to be able to use optimum generated ONNX models to use this tool, allowing many more models: e.g. BioBERT.

github-actions[bot] commented 3 months ago

:tada: This issue has been resolved in version 3.9.1 :tada:

The release is available on:

Your semantic-release bot :package::rocket: