Cannot try due large forces

[raimis]

According to the description, SPICE.hdf5 is downloaded from https://github.com/openmm/spice-dataset/releases:

spice-models/five-et/README.md

Lines 3 to 4 in b9023c7

	The script `createSpiceDataset.py` converts the dataset file SPICE.hdf5 downloaded from https://github.com/openmm/spice-dataset/releases
	to the format used by [TorchMD-Net](https://github.com/torchmd/torchmd-net). It generates a new file SPICE-processed.hdf5

The file contains samples with large force (openmm/spice-dataset#32) and createSpiceDataset.py doesn't filter them.

When I try to train the loss just fluctuates ~1e7 without any improvement.

[raimis]

@peastman could you clarify the force cutoff? And how did you apply it?

[peastman]

The filtering was implemented in openmm/spice-dataset#41, which was merged after that release was created. Once we're agreed on all remaining changes (especially anything for openmm/spice-dataset#42), I'll create a 1.1 release.

[raimis]

@peastman the cutoff in openmm/spice-dataset#41 is 1 Hartree/Bohr (≈ 49132 kJ/mol/nm), while in the email you mentioned 20000 kJ/mol/nm. Which ones was used?

[peastman]

My original model described in the first draft of the paper used 20,000 kJ/mol/nm. The five newer models described in the current draft used 1 hartree/bohr.