feat(pt): DPA-2 repinit compress (#4329)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

- **New Features**
- Introduced model compression functionality in the descriptor class,
allowing users to enable compression with specific parameters.
- Enhanced handling of serialized data for different descriptor types,
improving flexibility and efficiency.

- **Tests**
- Added a new test suite for the descriptor class, ensuring robust
testing of functionality with various configurations and floating-point
precisions.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Signed-off-by: Jinzhe Zeng <[email protected]>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>

deepmd/pt/model/descriptor/dpa2.py

@@ -31,10 +31,14 @@
     build_multiple_neighbor_list,
     get_multiple_nlist_key,
 )
+from deepmd.pt.utils.tabulate import (
+    DPTabulate,
+)
 from deepmd.pt.utils.update_sel import (
     UpdateSel,
 )
 from deepmd.pt.utils.utils import (
+    ActivationFn,
     to_numpy_array,
 )
 from deepmd.utils.data_system import (
@@ -306,6 +310,7 @@ def init_subclass_params(sub_data, sub_class):
         # set trainable
         for param in self.parameters():
             param.requires_grad = trainable
+        self.compress = False
 
     def get_rcut(self) -> float:
         """Returns the cut-off radius."""
@@ -859,3 +864,85 @@ def update_sel(
         )
         local_jdata_cpy["repformer"]["nsel"] = repformer_sel[0]
         return local_jdata_cpy, min_nbor_dist
+
+    def enable_compression(
+        self,
+        min_nbor_dist: float,
+        table_extrapolate: float = 5,
+        table_stride_1: float = 0.01,
+        table_stride_2: float = 0.1,
+        check_frequency: int = -1,
+    ) -> None:
+        """Receive the statistics (distance, max_nbor_size and env_mat_range) of the training data.
+
+        Parameters
+        ----------
+        min_nbor_dist
+            The nearest distance between atoms
+        table_extrapolate
+            The scale of model extrapolation
+        table_stride_1
+            The uniform stride of the first table
+        table_stride_2
+            The uniform stride of the second table
+        check_frequency
+            The overflow check frequency
+        """
+        # do some checks before the mocel compression process
+        if self.compress:
+            raise ValueError("Compression is already enabled.")
+        assert (
+            not self.repinit.resnet_dt
+        ), "Model compression error: repinit resnet_dt must be false!"
+        for tt in self.repinit.exclude_types:
+            if (tt[0] not in range(self.repinit.ntypes)) or (
+                tt[1] not in range(self.repinit.ntypes)
+            ):
+                raise RuntimeError(
+                    "Repinit exclude types"
+                    + str(tt)
+                    + " must within the number of atomic types "
+                    + str(self.repinit.ntypes)
+                    + "!"
+                )
+        if (
+            self.repinit.ntypes * self.repinit.ntypes - len(self.repinit.exclude_types)
+            == 0
+        ):
+            raise RuntimeError(
+                "Repinit empty embedding-nets are not supported in model compression!"
+            )
+
+        if self.repinit.attn_layer != 0:
+            raise RuntimeError(
+                "Cannot compress model when repinit attention layer is not 0."
+            )
+
+        if self.repinit.tebd_input_mode != "strip":
+            raise RuntimeError(
+                "Cannot compress model when repinit tebd_input_mode == 'concat'"
+            )
+
+        # repinit doesn't have a serialize method
+        data = self.serialize()
+        self.table = DPTabulate(
+            self,
+            data["repinit_args"]["neuron"],
+            data["repinit_args"]["type_one_side"],
+            data["exclude_types"],
+            ActivationFn(data["repinit_args"]["activation_function"]),
+        )
+        self.table_config = [
+            table_extrapolate,
+            table_stride_1,
+            table_stride_2,
+            check_frequency,
+        ]
+        self.lower, self.upper = self.table.build(
+            min_nbor_dist, table_extrapolate, table_stride_1, table_stride_2
+        )
+
+        self.repinit.enable_compression(
+            self.table.data, self.table_config, self.lower, self.upper
+        )
+        self.compress = True

deepmd/pt/utils/tabulate.py

@@ -95,11 +95,14 @@ def __init__(
             raise RuntimeError("Unknown activation function type!")
 
         self.activation_fn = activation_fn
-        self.davg = self.descrpt.serialize()["@variables"]["davg"]
-        self.dstd = self.descrpt.serialize()["@variables"]["dstd"]
-        self.ntypes = self.descrpt.get_ntypes()
+        serialized = self.descrpt.serialize()
+        if isinstance(self.descrpt, deepmd.pt.model.descriptor.DescrptDPA2):
+            serialized = serialized["repinit_variable"]
+        self.davg = serialized["@variables"]["davg"]
+        self.dstd = serialized["@variables"]["dstd"]
+        self.embedding_net_nodes = serialized["embeddings"]["networks"]
 
-        self.embedding_net_nodes = self.descrpt.serialize()["embeddings"]["networks"]
+        self.ntypes = self.descrpt.get_ntypes()
 
         self.layer_size = self._get_layer_size()
         self.table_size = self._get_table_size()
@@ -291,7 +294,13 @@ def _layer_1(self, x, w, b):
         return t, self.activation_fn(torch.matmul(x, w) + b) + t
 
     def _get_descrpt_type(self):
-        if isinstance(self.descrpt, deepmd.pt.model.descriptor.DescrptDPA1):
+        if isinstance(
+            self.descrpt,
+            (
+                deepmd.pt.model.descriptor.DescrptDPA1,
+                deepmd.pt.model.descriptor.DescrptDPA2,
+            ),
+        ):
             return "Atten"
         elif isinstance(self.descrpt, deepmd.pt.model.descriptor.DescrptSeA):
             return "A"

source/tests/pt/model/test_compressed_descriptor_dpa2.py

@@ -0,0 +1,149 @@
+# SPDX-License-Identifier: LGPL-3.0-or-later
+import unittest
+from typing import (
+    Any,
+)
+
+import numpy as np
+import torch
+
+from deepmd.dpmodel.descriptor.dpa2 import (
+    RepformerArgs,
+    RepinitArgs,
+)
+from deepmd.env import (
+    GLOBAL_NP_FLOAT_PRECISION,
+)
+from deepmd.pt.model.descriptor.dpa2 import (
+    DescrptDPA2,
+)
+from deepmd.pt.utils.env import DEVICE as PT_DEVICE
+from deepmd.pt.utils.nlist import build_neighbor_list as build_neighbor_list_pt
+from deepmd.pt.utils.nlist import (
+    extend_coord_with_ghosts as extend_coord_with_ghosts_pt,
+)
+
+from ...consistent.common import (
+    parameterized,
+)
+
+
+def eval_pt_descriptor(
+    pt_obj: Any, natoms, coords, atype, box, mixed_types: bool = False
+) -> Any:
+    ext_coords, ext_atype, mapping = extend_coord_with_ghosts_pt(
+        torch.from_numpy(coords).to(PT_DEVICE).reshape(1, -1, 3),
+        torch.from_numpy(atype).to(PT_DEVICE).reshape(1, -1),
+        torch.from_numpy(box).to(PT_DEVICE).reshape(1, 3, 3),
+        pt_obj.get_rcut(),
+    )
+    nlist = build_neighbor_list_pt(
+        ext_coords,
+        ext_atype,
+        natoms[0],
+        pt_obj.get_rcut(),
+        pt_obj.get_sel(),
+        distinguish_types=(not mixed_types),
+    )
+    result, _, _, _, _ = pt_obj(ext_coords, ext_atype, nlist, mapping=mapping)
+    return result
+
+
+@parameterized(("float32", "float64"), (True, False))
+class TestDescriptorDPA2(unittest.TestCase):
+    def setUp(self):
+        (self.dtype, self.type_one_side) = self.param
+        if self.dtype == "float32":
+            self.skipTest("FP32 has bugs:")
+            # ../../../../deepmd/pt/model/descriptor/repformer_layer.py:521: in forward
+            # torch.matmul(attnw.unsqueeze(-2), gg1v).squeeze(-2).view(nb, nloc, nh * ni)
+            # E       RuntimeError: expected scalar type Float but found Double
+        if self.dtype == "float32":
+            self.atol = 1e-5
+        elif self.dtype == "float64":
+            self.atol = 1e-10
+        self.seed = 21
+        self.sel = [10]
+        self.rcut_smth = 5.80
+        self.rcut = 6.00
+        self.neuron = [6, 12, 24]
+        self.axis_neuron = 3
+        self.ntypes = 2
+        self.coords = np.array(
+            [
+                12.83,
+                2.56,
+                2.18,
+                12.09,
+                2.87,
+                2.74,
+                00.25,
+                3.32,
+                1.68,
+                3.36,
+                3.00,
+                1.81,
+                3.51,
+                2.51,
+                2.60,
+                4.27,
+                3.22,
+                1.56,
+            ],
+            dtype=GLOBAL_NP_FLOAT_PRECISION,
+        )
+        self.atype = np.array([0, 1, 1, 0, 1, 1], dtype=np.int32)
+        self.box = np.array(
+            [13.0, 0.0, 0.0, 0.0, 13.0, 0.0, 0.0, 0.0, 13.0],
+            dtype=GLOBAL_NP_FLOAT_PRECISION,
+        )
+        self.natoms = np.array([6, 6, 2, 4], dtype=np.int32)
+
+        repinit = RepinitArgs(
+            rcut=self.rcut,
+            rcut_smth=self.rcut_smth,
+            nsel=10,
+            tebd_input_mode="strip",
+            type_one_side=self.type_one_side,
+        )
+        repformer = RepformerArgs(
+            rcut=self.rcut - 1,
+            rcut_smth=self.rcut_smth - 1,
+            nsel=9,
+        )
+
+        self.descriptor = DescrptDPA2(
+            ntypes=self.ntypes,
+            repinit=repinit,
+            repformer=repformer,
+            precision=self.dtype,
+        )
+
+    def test_compressed_forward(self):
+        result_pt = eval_pt_descriptor(
+            self.descriptor,
+            self.natoms,
+            self.coords,
+            self.atype,
+            self.box,
+        )
+        self.descriptor.enable_compression(0.5)
+        result_pt_compressed = eval_pt_descriptor(
+            self.descriptor,
+            self.natoms,
+            self.coords,
+            self.atype,
+            self.box,
+        )
+
+        self.assertEqual(result_pt.shape, result_pt_compressed.shape)
+        torch.testing.assert_close(
+            result_pt,
+            result_pt_compressed,
+            atol=self.atol,
+            rtol=self.atol,
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()