Hi does haystack only support fastembed for sparse (doc/text) ?

[superbaim]

Hi,

Does haystack(2.x 3 June 2024) only support fastembed for sparse embed(doc/text) ?
And One more thing for sparse embbed fastembed it doesnt support CUDA ?

Thanks

[anakin87]

Hello!

Can you clarify what you mean by "sparse embed(doc/text)"?

FastEmbed recently added support for GPU, but this is not currently supported in the Haystack-FastEmbed integration.

[superbaim]

Hi,

Thanks for second question answer,
So Sorry, im currently exploring hybrid retrieval(sparse+dense), for dense i see many option for dense embedding like sentence transformer and huggingface, but for sparse i only see Fastembedd(FastembedSparseTextEmbedder+FastembedSparseDocumentEmbedder) and the model that fastembed support, am i right ?

Thanks

[anakin87]

Yes, you are right. Have you any other ideas?