MingxiaoLi

> @Mingxiao-Li Can we close this issue? Yes, many thanks.

> [The stochastic sampler](https://github.com/openai/consistency_models/blob/6d26080c58244555c031dbc63080c0961af74200/cm/karras_diffusion.py#L657) is the same, assuming you finish on `t_min`. In this case the scaling applied to the noise will be zero, so it finishes with the final...

> Been a while since I've been in depth with this code, so this may be a naive question. Are you sure `ts` should be in ascending order? Having the...

> > > > Been a while since I've been in depth with this code, so this may be a naive question. Are you sure `ts` should be in ascending...

> > > > Been a while since I've been in depth with this code, so this may be a naive question. Are you sure `ts` should be in ascending...

> We trained somewhere between 130-150 epochs. > How much data did you generate and did you generate episodes from all training scenes? > Which method were you trying to...

> I am sorry. I don't get it. Do you mean I can not start with training the CM2 ? But I think there is no connections between CM2 and...

> You can technically start with CM2 as it is independent of CM2-GT. I only suggested to start with CM2-GT because it trains much faster and is easier to debug....

I also found another problem in the code. In the data loader, in line 222,223 and 727 728 I found the below comment and code : comment: get the relative...

> Yes, please add pose noise when generating data for CM2 model. > > That is not a problem in the code. The comment above is outdated. The "pose_coords" is...