François Bérenger comments

Results 420 comments of


                                            François Bérenger

List of supported atoms nowhere to be seen in the documentation [documentation enhancement]

From the code ([torchani/data/__init__.py](https://github.com/aiqm/torchani/blob/40cf334dbdf71903f30be8560784ca793b830221/torchani/data/__init__.py#L157)), I suspect this: CHNOSFCl

List of supported atoms nowhere to be seen in the documentation [documentation enhancement]

From the paper, also CHNOSFCl.

List of supported atoms nowhere to be seen in the documentation [documentation enhancement]

Thanks, I was right then. This is a key thing; it should be clearly stated in the documentation too.

GPU-enabled ANI-2x potential for MD simulations in ASE

In bash shell: ```bash function gpu-watch () { watch nvidia-smi -a --display=utilization } ```

faster rmsd calculations in rdkit

Here is some code that I have used in production: https://github.com/UnixJunkie/durandal_qcp/blob/master/src/qcprot.cc look for rmsd_without_rotation_matrix. I think that's what most people want. When you want to superpose molecules, that's another business.

new file: bin/mols2html.py

related to https://github.com/cbouy/mols2grid/issues/53

Define number of cores to be used does not work

I confirm that the env. var OMP_NUM_THREADS=1 is being royally ignored

Define number of cores to be used does not work

Accelerating training is the goal. OMP_NUM_THREADS is a very simple way to limit the number of cores when running jobs in a cluster.

Making Predictions on Large Datasets

Would that work: load one shard of your dataset at a time (let's say 50k molecules), predict only on one shard at a time.

Totally different got similarity of 1

Do you have other such molecular pairs?