Bruno Sánchez-Andrade Nuño

More specifically, my issue is to create a stackstac that reads overviews, so I can for example create a median value composite over a country-size roi reading overviews.

Thanks everyone for the digging. In the meantime, for reference of others who need a workaround. It seems that [ODC-STAC](https://odc-stac.readthedocs.io/en/latest/) can handle these cases reading overviews ```python import odc.stac,odc import...

This also fixes the export error for the missing `public/data` folder.  related: #213.

Hi Rob!! I think the right move here is to copy a representative sample of embeddings to source.coop I don't think if it makes sense to publicly host a copy...

Thanks for creating this Issue @jaanli ! This sounds very interesting and please do keep us in the loop of progress. I'd say pass on patch-level embeddings. As I explain...

I chatted with @yellowcap and @lukaskondmann and I think I was wrong. > If you don't mask enough, the task is too easy for the MAE and it will not...

Tagging here something I [learned today](https://github.com/Clay-foundation/model/issues/127#issuecomment-1898541474), which is that for each self-attention patch, the 13 layers are grouped into [6 groups](https://github.com/Clay-foundation/model/blob/main/src/model_clay.py#L584), and we create one embedding per group. I still...

I keep coming up with the need to do this. @yellowcap and I had a great conversation where we decided NOT to do this for now. The reason being that...

Bumping this up again. Specially as we move away from fixed bands and band groups, and we focus on e.g. similarly search at the patch level, it seems critical that...

Talking with @yellowcap, it seems we are already merging all bands into a single patch embedding. @srmsoumya to confirm and close here.