MiDi
MiDi copied to clipboard
cvignac
MiDi: Mixed Graph and 3D Denoising Diffusion for Molecule Generation
Hi Clément, Hope you are well. In the paper you mentioned that the adaptive version is trained form a uniform checkpoint. Is that 500 epochs to train uniform and another...
Hi, Can you provide the data indices for your original split of GEOM? I find some issues in the GEOM dataset and I need to re-process it. I want to...
I want to retrain MIDI on the Chembl with explicit hydrogens. During the training process, I found that after more than 100 epochs of training, the validity of the model...
**Description** The line: ```python edge_types[edge_types == 4] = 1.5 ``` fails because edge_types is a long tensor, truncating 1.5 to 1. This replaces 4 with 1 instead of 1.5, affecting...
Congrats on this very nice work! I find that there are no files about GEOM Datasets of train validation test in the URL. Could you please upload them again or...
First and foremost, I would like to express my admiration for the work all parties involved, have done with both DiGress and here MiDi. It is very impressive and I...
Hi Clément, When trying to use the testing procedure, I encountered a bug related to input_dims. It seems that in the current version of the code, they are updated twice...
When setting up the environment with conda, pyg can not be installed. To solve this, when setting up the environment specify the RDKit version: `conda create -c conda-forge -n MoleculeDiffusion...
Hello, I'm trying to use MiDi to generate molecules based on the model trained on GEOM with explicit H. The trained model requires the dataset_infos as input, which needs the...
