deepmd.dpmodel.model.unimol_pretrain_model

deepmd.dpmodel.model.unimol_pretrain_model#

Model wrapper for Uni-Mol v1 self-supervised pretraining.

Attributes#

DPUniMolPretrainModel_

Classes#

UniMolPretrainModel

Uni-Mol v1 molecular pretraining.

Functions#

_reject_periodic(→ None)

Refuse a periodic cell, with an explanation rather than an allocation.

Module Contents#

deepmd.dpmodel.model.unimol_pretrain_model._reject_periodic(box) → None[source]#: Refuse a periodic cell, with an explanation rather than an allocation.

deepmd.dpmodel.model.unimol_pretrain_model.DPUniMolPretrainModel_[source]#

class deepmd.dpmodel.model.unimol_pretrain_model.UniMolPretrainModel(*args: Any, **kwargs: Any)[source]#

Bases: deepmd.dpmodel.model.dp_model.DPModelCommon, DPUniMolPretrainModel_

Uni-Mol v1 molecular pretraining.

Predicts the element of every corrupted atom, denoises the coordinates and predicts the clean pairwise distances, and reports the two norm regularisers of the backbone. Nothing here reduces to a frame total and nothing is differentiated with respect to the coordinates: this is representation learning, not a potential energy surface.

call(coord: deepmd.dpmodel.array_api.Array, atype: deepmd.dpmodel.array_api.Array, box: deepmd.dpmodel.array_api.Array | None = None, fparam: deepmd.dpmodel.array_api.Array | None = None, aparam: deepmd.dpmodel.array_api.Array | None = None, do_atomic_virial: bool = False, charge_spin: deepmd.dpmodel.array_api.Array | None = None) → dict[str, deepmd.dpmodel.array_api.Array][source]#

Evaluate the pretraining heads on a frame.

Raises:

ValueError: If a periodic cell is supplied. Uni-Mol is molecular: it has no cut-off, so a cell would ask the neighbour-list builder for an astronomical number of images before any other check could fire.

call_lower(extended_coord: deepmd.dpmodel.array_api.Array, extended_atype: deepmd.dpmodel.array_api.Array, nlist: deepmd.dpmodel.array_api.Array, mapping: deepmd.dpmodel.array_api.Array | None = None, fparam: deepmd.dpmodel.array_api.Array | None = None, aparam: deepmd.dpmodel.array_api.Array | None = None, do_atomic_virial: bool = False, charge_spin: deepmd.dpmodel.array_api.Array | None = None) → dict[str, deepmd.dpmodel.array_api.Array][source]#: Evaluate the pretraining heads on an extended frame.

translated_output_def() → dict[str, deepmd.dpmodel.output_def.OutputVariableDef][source]#: The head outputs, passed through under their own names.