EiffL / sfh-inference-hackathon

Repository for SFH inference hackathon at AstroInfo 2021
MIT License
7 stars 2 forks source link

Setting up clean notebook with generative modeling to use `tng100` dataset #28

Open EiffL opened 2 years ago

EiffL commented 2 years ago

Let's use this issue to talk about what we want to see in the first "clean" notebook.

EiffL commented 2 years ago

This is one the last notebooks that I had I think: https://github.com/EiffL/sfh-inference-hackathon/blob/main/notebooks/PixelCNN_experiments_EiffL.ipynb

Did you guys have other notebooks? It'd be good to list them all to see what we had and combine into one baseline model and notebook, from which we can then build several variants.

laurilaatu commented 2 years ago

To condition with images we would need to replace sed_net with network using Conv2D, maybe tf.keras.applications.resnet50 could be a good candidate, as well as implement some preprocessing to the images? Is there anything beyond this to be done?