Clay-foundation
Add tutorial for creating embeddings over AOI
Show how to use the clay torchscript encoder to create embeddings over AOI.
Hey @yellowcap, I'm getting:
`--------------------------------------------------------------------------- ValueError Traceback (most recent call last) Cell In[19], line 147 137 pixels = transform(pixels) 139 datacube = ( 140 torch.tensor(pixels, dtype=torch.float32, device="cuda"), 141 torch.tensor(np.hstack((week_norm, hour_norm)), dtype=torch.float32, device="cuda"), (...) 144 torch.tensor(gsd, dtype=torch.float32, device="cuda"), 145 ) --> 147 clay_encoder = torch.export.load("clay-v1-encoder.pt").module() 149 # Run the clay encoder 150 with torch.no_grad():
ValueError: invalid literal for int() with base 10: b'5.1'`
when running the notebook. I tried changing to torch.load("clay-v1-encoder.pt").module(), but that didn't seem to help either. Any ideas?
@4242psherman4242 which pytorch version are you using to run the notebook?
@4242psherman4242 which pytorch version are you using to run the notebook?
Hey @srmsoumya this is using version 2.1.2 of pytorch
@4242psherman4242 Can you bump the pytorch version to ~2.3? torch.export is still under development & has breaking changes in minor versions.
Hi @yellowcap is it possible to resolve the aforementioned conflicts soon? Would be great to get this super clean way of running inference merged
