hau

updating the repo?

5

is there some way to easy pull new updates / new components that are added to the library? since it's not released as a package

Supported context window length for each model?

12

what's the supported context window length for each model?

speeding up inference

40

- stateful load models once and then generate again - use torch compile?

for those with limited VRAM, any plans to support quantized versions of models?

tensorrt error

17

TypeError: streamdiffusion.acceleration.tensorrt.compile_unet() got multiple values for keyword argument 'opt_batch_size' ?

Change the default font

1

Can the default font by changed to Geist Sans?

Would it be possible to export a nn.Module that uses odeint under the hood to ONNX?

webgpu?

3

do you think switching the engine to webgpu would help with performance? babylon is supposed to support webgpu out of the box