CAPI follow-up experiments

[utxrsh]

#16 this PR adds an initial setup for CAPI and some baseline results. Next goal is to try and match up the suite of experiments designed for the MAE model to have a fairer comparison. (Noting some to-dos, anyone with more ideas feel free to add to this list)

• implement the spatiotemporal version of the model (high effort)
  • replicate the temporal_patch_size = 16 experiment first to compare results against the CAPI baseline
• add PCA feature visualisations similar to MAE (medium effort)
  • compare feature maps across CAPI and MAE

[sagarigrandhi]

I worked on the PCA visualizations for MAE and can extend the same to CAPI. Let me know if you’d like me to do it for the CAPI baseline.

[utxrsh]

I worked on the PCA visualizations for MAE and can extend the same to CAPI. Let me know if you’d like me to do it for the CAPI baseline.

yes sure, that'd be very helpful, thanks