Nunchaku model int4 probably doesn't work with RTX4070 with Ada Lovelace protocol

[dusatrnavac-ai]

Hi, I have an issue with the nunchaku model svdq-int4_r32-flux.1-krea-dev,safetensors that obviously won't work with my ASUS DUAL-RTX4070-O12G graphic card.
Since my graphic card uses Ada Lovelace protocol, maybe AI diffusion doesn't support it and that is the reason why i keep gettting the same error when using int4 nunchaku: "Disconnected from the server, trying to reconnect..."
And when i try to use fp4 nunchaku version it gives me an error message: "Server error: Please use "int4" quantization for Turing, Ampere and Ada GPUs.

So I wonder if there is any way to update AI diffusion with Ada Lovelace protocol for RTX4070 graphics? Could this be the reason for breaking out the connection with the server?

[lokpro]

https://docs.interstice.cloud/common-issues/
log files?

client.log
server.log