Nicolas Renaud

The function `_get_relevance` of `NeuralNet` (that is called at each epoch) is really slow and takes up to a few hours on cartesius when training on 001-003 of BM5. In...

When loading a model that has been trained on GPU and loading it on CPU the line 230-240 of NeuralNet.py removes the state_dict keys and therefore we can't load the...

Redo Figure 2

4

We need a better representation of the crystal and the two types of interface for figure 2. Maybe a schematic of crystal versus biological would be good since we targeting...

Redo Figure 3

3

I think the histogram and the results should be redrawn. For the results it might be best to only plot the results on the test set for different scenario and...

Software Architecture Picture

2

Create a picture that represents the overall architecture. Put it in supp matt

Go through the text and polish it

Data Generation picture

1

Create a picture that illustrate the parallel data generation. Put it in supp mat. Add a figure with computation time (generatinng a grid, aligning a pdb file and the size...

Organize Supplementary Material

1

The supp mat should be in a separate file (right ?). We have to decide what we put there and organize the content first. Create the table of content and...

Go through the text and polish it

Go through the text and polish it