Spaces:

captain-awesome

/

Docuverse-zephyr-beta

like 4

Sleeping

Thank you @MLconArtist just did and it works well now, only that it is a bit slower, of course due to the CPU, any waay to make the execution faster on a CPU, please do share.
Thanks again

I am glad to hear that it works.

The CPU part, even I am trying to figure that out. We kind of have like 2 options-

1. Use a small model, like a Lamini-Flan-T5-783M, which is fast and give coherent responses, but it falls apart when the query is a bit complex.
2. Use .gguf files of bigger models, the same thing that you are doing now. These give good responses but take a lot of time sometimes.

Another thing I have observed is that the huggingface spaces is actually slower than my pc.