Is it possible to deploy ViT using hls4ml?

[zjnyly]

Recently I am required to deploy a transformer model on FPGA using hls4ml by my mentor.

I am not sure if this is possible, because the model is too large and may need to buffer a large amont of intermediate data.

So, is it possible, or worth to try?

[ayusdixit]

same issue with me

[liuzrcc]

Same