Thanks for hosting me, Brad.

When we first started working on Diffusion Policy, which felt like an "unlock" for fitting human demonstration data, the explanation I gave myself was that "human demonstration data was subtly much more multimodal than we thought, so we need the modeling power of denoising diffusion to fit it properly." What Michael's paper brings to light is that some of this multimodality was probably exacerbated (more than I realized) by the fact that we've only been giving our models incredibly short observation context lengths. For me, this is pretty compelling evidence that we should be pushing for policies with long-context lengths and short action prediction horizons.

Thanks to @HeySamantha and @Bloomberg for covering our launch bloomberg.com/news/articles/…

When it comes to diffusion policies, VLAs, and LBMs, we all know that data quality matters. Many of us have spent considerable time filtering data, and discarding large amounts of data that might not be considered to be sufficiently high quality. This feels very inefficient. We’ve seen several proposals of late for how we might learn from suboptimal data. pi0.7, for instance, suggested that we learn models conditioned on the data quality. Adam’s new paper exposes a different (complementary) angle. When we’re training denoising diffusion models, each data point is potentially used several times — at different noise levels. **The value of a data point might be different depending on the noise level.** Adam’s work explores this nicely and shows that he can extract value even out of significant portions of the OXE dataset that many of us had previously discarded as suboptimal. Definitely worth having a look.

I've been saying for years that the biggest challenge for simulation in robotics is not actually the physics engine (although you do have to get that right). The real challenge is capturing the *diversity* of the real world. There was no doubt that generative AI had the potential to change that, but it's still amazing to see it take shape. Watching Nick's incredibly fast progress has convinced me that content generation might not actually be a bottleneck anymore. This is a beautiful combination of hardened tools for e.g. low-level mesh processing with the latest tools for generative asset creation, wrapped in a powerful agentic workflow. Please do give it a try and share your feedback.

Very proud of Nicholas, who recently shared scalable-real2sim.github.io/ (for physics-quality assets from a small amount of interaction with a robot) and is now following up with his work on scene-level generation.

Side note: I'm proud of the title of this paper, which we intentionally made pretty narrow/specific. I think that some of the most important work that we have to do as a field right now is careful empirical work to interrogate the properties of these models that we're creating.

One of the most interesting take-aways for me is that "high-performing policies need to know whether they are executing in sim or in real." A number of implications flow from that, including that sim+real cotraining can decrease performance if the visual gap is too small.