search

tummfm / difftre

Learning neural network potentials from experimental data via Differentiable Trajectory Reweighting
Apache License 2.0
28 stars 9 forks source link

Using pre-trained MLP in DiffTre #5

Closed longkunxuluke closed 6 months ago

longkunxuluke commented 6 months ago

Greetings. We are interested in using pre-trained MLP in DiffTre but we found there is no available example in the three notebooks you shared. In the paper we noticed you mentioned it would be a future work, I am wondering if you have more updated papers/demos showing how to use pre-trained MLP in DiffTre, or if you have any suggestions in this regard, e.g., which types of pre-trained MLP is supported with current DiffTre framework? Many thanks !

S-Roecken commented 6 months ago

Hi, we currently have a paper in review that employs a pre-trained MLP. For details see: https://www.researchgate.net/publication/374051787_Accurate_machine_learning_force_fields_via_experimental_and_simulation_data_fusion

Best Sebastien