Conversation
| from typing import Optional | ||
|
|
||
| import lobster.model.latent_generator.latent_generator as latent_generator | ||
| import sys |
There was a problem hiding this comment.
so that there aren't two cmdline folders in lobster, could you pls move the actual entry point to lobster/cmdline and keep the definitions like class LatentEncoderDecoder somewhere inside src/model/latent_generator. Then src/lobster/model/latent_generator/latent_generator/cmdline could be deleted
| # Ligand Models | ||
| # These models are optimized for ligand structure analysis | ||
| "LG Ligand 20A": { | ||
| "description": "Ligand only model with 20Å spatial attention", |
There was a problem hiding this comment.
I think these should be hydra configs that are loaded?
| description="Load a TokenizerMulti model and encode inputs", | ||
| formatter_class=argparse.RawDescriptionHelpFormatter, | ||
| epilog=""" | ||
| Available Models: |
There was a problem hiding this comment.
I think should be created on the fly based on the loaded configs to avoid having to keep it up to date manually?
| Main function for loading a model and encoding inputs. | ||
| """ | ||
| # Use argparse instead of hydra for better handling of paths with '=' in them | ||
| parser = argparse.ArgumentParser( |
There was a problem hiding this comment.
do we need all PDBs here or would just one be enough? to avoid having the repo too large. there is some PDB file in test_data already btw
|
LG is double nested |
There was a problem hiding this comment.
stuff inside cmdline seems a bit overloaded and includes some model definitions etc - would be good to split it up?
Description
adding latentgenerator code to lobster
Type of Change