A possible mistake in the rasterization cuda code?

[TeCai]

Hi, in rasteriza_indices_in_range.cu and rasterize_to_pixels_fwd.cu I noticed that while counting valid Gaussians,
https://github.com/nerfstudio-project/gsplat/blob/0880d2b471e6650d458aa09fe2b2834531f6e93b/gsplat/cuda/csrc/rasterize_to_indices_in_range.cu#L148-L153C14

This Gaussian is not counted when break is used, but it should be counted as it has a transmittance>1e-4. I referred to the original 3dgs code they used continue instead of break
https://github.com/graphdeco-inria/diff-gaussian-rasterization/blob/9c5c2028f6fbee2be239bc4c9421ff894fe4fbe0/cuda_rasterizer/forward.cu#L363C1-L369C1

Is this an error in the implementation or am I missing something?

Thanks!