Thanks for the great work! While running the T1 Force task, I noticed that in `utils/plotting_forces.py`, the RMSE is computed as follows: ``` forces_mse_xyz = (forces_pred - forces_gt) ** 2 forces_rmse_xyz = np.sqrt(forces_mse_xyz) * 1000 mean = f_rmse.mean() ``` I'm wondering whether the order of the square root and mean operations in the RMSE calculation might be incorrect. Shouldn't it instead be: ``` forces_mse_xyz = (forces_pred - forces_gt) ** 2 forces_rmse_xyz = np.sqrt(forces_mse_xyz.mean()) * 1000 ``` That is, computing the mean first and then taking the square root, rather than the other way around?
Thanks for the great work! While running the T1 Force task, I noticed that in
utils/plotting_forces.py, the RMSE is computed as follows:I'm wondering whether the order of the square root and mean operations in the RMSE calculation might be incorrect. Shouldn't it instead be:
That is, computing the mean first and then taking the square root, rather than the other way around?