Luis Pineda

@Muon2 did you notice any improvements with other optimization parameters?

Hi @CavalloneChen. Apologies for the long delay. Do you have a short repro script for this? We haven't tried using `torch.compile` yet, so I don't have a lot of insight...

Hi @Dhanushvarma. Sorry, we don't have this functionality yet.

Hi @richardrl. As a matter of fact, we are about to merge this functionality in #513, as part of our new [labs.lie](https://github.com/facebookresearch/theseus/blob/main/examples/lie_labs.py) package. Below is a simple example that uses...

Hi @FanWu-fan, can you post a full stack trace for the error? Regarding the `set_lie_tangent_enabled`, that context makes it so that when doing `optim.step()`, instead of simply adding the gradient...

Hi @FanWu-fan. The problem is that you are importing `SO3` from `th` instead of `lieF`; these are different objects. `th.SO3` is a `Variable` subclass, which wraps a tensor so that...

I have a lot of checkpoints stored for different versions of the library, but need to dig a little bit to check if any correspond to the latest version. I...

Hi @xphsean12. Are you referring the the numeric jacobians? We haven't made this optimization yet, but note that this function is only used in our unit tests. For users the...

Would [this](https://github.com/facebookresearch/theseus/blob/main/tests/embodied/measurements/test_moving_frame_between.py#L47) be useful for you? It's an example of how we test our analytic jacobians numerically in our unit tests. You can follow a similar template to verify the...

Hi @DanielTakeshi, thanks for posting the updated info! For completeness, I think closing the issue would be best after migrating all cost function tests to use the new utility, because...