Spaces:

microsoft

/

Promptist

like 310

Running

I think that'd be super cool! What do you think @unilm ?

I think you should pull my changes in first, then decide if you need a GPU.

I think you should pull my changes in first, then decide if you need a GPU.

I duplicate the demo at https://huggingface.co/spaces/unilm/Promptist-faster and merge your update.

I also added a note as follows:

Note: This is a version with beam_size=1 while the original demo uses beam_size=8. So there would be a difference in terms of performance, but this demo is much faster. Many thanks to @HughPH for pointing out this improvement.

Hey there! I'm Omar from Hugging Face. Would you be interested in receiving a free, temporary GPU grant? It would allow the demo to be much faster. Some parts of the code might need to change to have the model/inputs on the GPU.

We didn't have plans to host the demo using GPUs now. It's supposed to load the models locally using the huggingface library. The demo shows an example of how to load the checkpoint. I appreciate your kindness in the GPU grant.