corvic-engine 0.3.0rc85__tar.gz → 0.3.0rc87__tar.gz

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/Cargo.lock

@@ -10,13 +10,13 @@ checksum = "ace50bade8e6234aa140d9a2f552bbee1db4d353f69b8217bc503490fc1a9f26"
 
 [[package]]
 name = "cfg-if"
-version = "1.0.0"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
+checksum = "9555578bc9e57714c812a1f84e4fc5b4d21fcb063490c624de019f7464c91268"
 
 [[package]]
 name = "corvic-engine"
-version = "0.3.0-rc.85"
+version = "0.3.0-rc.87"
 dependencies = [
  "numpy",
  "pyo3",
@@ -57,7 +57,7 @@ checksum = "48c757948c5ede0e46177b7add2e67155f70e33c07fea8284df6576da70b3719"
 
 [[package]]
 name = "engine-walk"
-version = "0.3.0-rc.85"
+version = "0.3.0-rc.87"
 dependencies = [
  "version_check",
 ]
@@ -293,9 +293,9 @@ checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d"
 
 [[package]]
 name = "syn"
-version = "2.0.101"
+version = "2.0.102"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ce2b7fc941b3a24138a0a7cf8e858bfc6a992e7978a068a5c760deb0ed43caf"
+checksum = "f6397daf94fa90f058bd0fd88429dd9e5738999cca8d701813c80723add80462"
 dependencies = [
  "proc-macro2",
  "quote",

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/Cargo.toml

@@ -3,7 +3,7 @@ resolver = "2"
 members = ["python"]
 
 [workspace.package]
-version = "0.3.0-rc.85"
+version = "0.3.0-rc.87"
 edition = "2021"
 publish = false
 

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: corvic-engine
-Version: 0.3.0rc85
+Version: 0.3.0rc87
 Classifier: Environment :: Console
 Classifier: License :: Other/Proprietary License
 Classifier: Programming Language :: Python :: Implementation :: CPython

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/pyproject.toml

@@ -18,7 +18,7 @@ authors = [
 ]
 readme = "README.md"
 description = "Seamless embedding generation and retrieval."
-version = "0.3.0-rc.85" # sync this with tool.poetry.version below
+version = "0.3.0-rc.87" # sync this with tool.poetry.version below
 
 # TODO(ddn): Pull dependencies from tool.poetry.dependencies. We use maturin as
 # the build system but poetry for the development environment.
@@ -73,7 +73,7 @@ packages = [
 ]
 description = ""
 authors = []
-version = "0.3.0-rc.85"
+version = "0.3.0-rc.87"
 
 [tool.poetry.scripts]
 check = "corvic_check.cli:main"

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_base_model.py

@@ -3,7 +3,6 @@ from collections.abc import Callable, Iterable
 from typing import Generic
 
 import sqlalchemy as sa
-import sqlalchemy.orm as sa_orm
 
 from corvic import eorm, system, transfer
 from corvic.emodel._proto_orm_convert import (
@@ -59,7 +58,7 @@ class OrmBackedModel(
             [sa.Select[tuple[transfer.OrmT]]], sa.Select[tuple[transfer.OrmT]]
         ]
         | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[transfer.ProtoT]] | NotFoundError | InvalidArgumentError:
         return super().list_as_proto(
             client,
@@ -91,7 +90,7 @@ class HasIdOrmBackedModel(
             [sa.Select[tuple[transfer.OrmHasIdT]]], sa.Select[tuple[transfer.OrmHasIdT]]
         ]
         | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[transfer.ProtoHasIdT]] | NotFoundError | InvalidArgumentError:
         return super().list_as_proto(
             client,

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_completion_model.py

@@ -7,8 +7,6 @@ import datetime
 from collections.abc import Iterable, Sequence
 from typing import Literal, TypeAlias
 
-from sqlalchemy import orm as sa_orm
-
 from corvic import eorm, system, transfer
 from corvic.emodel._base_model import OrgWideStandardModel
 from corvic.emodel._defaults import Defaults
@@ -151,7 +149,7 @@ class CompletionModel(
         created_before: datetime.datetime | None = None,
         client: system.Client | None = None,
         ids: Iterable[CompletionModelID] | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[CompletionModel]] | NotFoundError | InvalidArgumentError:
         """List completion models."""
         client = client or Defaults.get_default_client()
@@ -179,7 +177,7 @@ class CompletionModel(
         cls,
         completion_model_id: CompletionModelID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[CompletionModel] | NotFoundError:
         client = client or Defaults.get_default_client()
         return cls.load_proto_for(completion_model_id, client, session).map(

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_feature_view.py

@@ -691,7 +691,7 @@ class FeatureView(
         cls,
         feature_view_id: FeatureViewID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[Self] | NotFoundError:
         client = client or Defaults.get_default_client()
         return cls.load_proto_for(feature_view_id, client, session).map(
@@ -723,7 +723,7 @@ class FeatureView(
         created_before: datetime.datetime | None = None,
         client: system.Client | None = None,
         ids: Iterable[FeatureViewID] | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[FeatureView]] | NotFoundError | InvalidArgumentError:
         """List resources."""
         client = client or Defaults.get_default_client()

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_pipeline.py

@@ -100,7 +100,7 @@ class Pipeline(StandardModel[PipelineID, models_pb2.Pipeline, eorm.Pipeline]):
         cls,
         obj_id: PipelineID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[SpecificPipeline] | NotFoundError:
         client = client or Defaults.get_default_client()
         match cls.list_as_proto(
@@ -134,7 +134,7 @@ class Pipeline(StandardModel[PipelineID, models_pb2.Pipeline, eorm.Pipeline]):
         created_before: datetime.datetime | None = None,
         client: system.Client | None = None,
         ids: Iterable[PipelineID] | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[SpecificPipeline]] | InvalidArgumentError | NotFoundError:
         client = client or Defaults.get_default_client()
 

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_proto_orm_convert.py

@@ -231,7 +231,7 @@ def add_orm_org_mixin_to_session(
     orm_obj: OrmHasIdBelongsToOrgT,
     proto_obj: _ModelHasIdBelongsToOrgProto,
     id_class: type[transfer.OrmIdT],
-    session: sa_orm.Session,
+    session: eorm.Session,
 ) -> Ok[OrmHasIdBelongsToOrgT] | orm.InvalidORMIdentifierError:
     match transfer.translate_orm_id(proto_obj.id, id_class):
         case Ok(orm_id):
@@ -257,7 +257,7 @@ def add_orm_room_mixin_to_session(
     orm_obj: OrmHasIdBelongsToRoomT,
     proto_obj: _ModelHasIdBelongsToRoomProto,
     id_class: type[transfer.OrmIdT],
-    session: sa_orm.Session,
+    session: eorm.Session,
 ) -> Ok[OrmHasIdBelongsToRoomT] | orm.InvalidORMIdentifierError:
     room_id = eorm.RoomID(proto_obj.room_id)
     match room_id.to_db():
@@ -270,7 +270,7 @@ def add_orm_room_mixin_to_session(
 
 
 def _resource_pipeline_to_orm(
-    proto_obj: models_pb2.Resource, orm_obj: eorm.Resource, session: sa_orm.Session
+    proto_obj: models_pb2.Resource, orm_obj: eorm.Resource, session: eorm.Session
 ) -> Ok[None] | InvalidArgumentError:
     if proto_obj.pipeline_id:
         match transfer.translate_orm_id(proto_obj.pipeline_id, eorm.PipelineID):
@@ -296,7 +296,7 @@ def _resource_pipeline_to_orm(
 
 
 def resource_proto_to_orm(
-    proto_obj: models_pb2.Resource, session: sa_orm.Session
+    proto_obj: models_pb2.Resource, session: eorm.Session
 ) -> Ok[eorm.Resource] | InvalidArgumentError:
     orm_obj = eorm.Resource(
         name=proto_obj.name,
@@ -321,11 +321,11 @@ def resource_proto_to_orm(
 def _ensure_id(
     proto_obj: transfer.ProtoHasIdT,
     proto_to_orm: Callable[
-        [transfer.ProtoHasIdT, sa_orm.Session],
+        [transfer.ProtoHasIdT, eorm.Session],
         Ok[transfer.OrmHasIdT] | InvalidArgumentError,
     ],
     id_type: type[transfer.OrmIdT],
-    session: sa_orm.Session,
+    session: eorm.Session,
 ) -> Ok[transfer.OrmIdT] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     match transfer.translate_orm_id(proto_obj.id, id_type):
         case orm.InvalidORMIdentifierError() as err:
@@ -344,7 +344,7 @@ def _ensure_id(
 
 
 def pipeline_proto_to_orm(  # noqa: C901
-    proto_obj: models_pb2.Pipeline, session: sa_orm.Session
+    proto_obj: models_pb2.Pipeline, session: eorm.Session
 ) -> Ok[eorm.Pipeline] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     orm_obj = eorm.Pipeline(
         name=proto_obj.name,
@@ -389,7 +389,7 @@ def pipeline_proto_to_orm(  # noqa: C901
 
 
 def source_proto_to_orm(
-    proto_obj: models_pb2.Source, session: sa_orm.Session
+    proto_obj: models_pb2.Source, session: eorm.Session
 ) -> Ok[eorm.Source] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     orm_obj = eorm.Source(
         name=proto_obj.name,
@@ -399,7 +399,7 @@ def source_proto_to_orm(
 
 
 def space_proto_to_orm(
-    proto_obj: models_pb2.Space, session: sa_orm.Session
+    proto_obj: models_pb2.Space, session: eorm.Session
 ) -> Ok[eorm.Space] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     match _ensure_id(
         proto_obj.feature_view, feature_view_proto_to_orm, eorm.FeatureViewID, session
@@ -423,7 +423,7 @@ def space_proto_to_orm(
 
 
 def feature_view_proto_to_orm(
-    proto_obj: models_pb2.FeatureView, session: sa_orm.Session
+    proto_obj: models_pb2.FeatureView, session: eorm.Session
 ) -> Ok[eorm.FeatureView] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     orm_obj = eorm.FeatureView(
         name=proto_obj.name,
@@ -493,7 +493,7 @@ def _feature_view_source_proto_to_orm(
     proto_obj: models_pb2.FeatureViewSource,
     room_id: eorm.RoomID,
     feature_view_id: eorm.FeatureViewID,
-    session: sa_orm.Session,
+    session: eorm.Session,
 ) -> Ok[eorm.FeatureViewSource] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     match _ensure_id(proto_obj.source, source_proto_to_orm, eorm.SourceID, session):
         case orm.InvalidORMIdentifierError() | InvalidArgumentError() as err:
@@ -514,14 +514,14 @@ def _feature_view_source_proto_to_orm(
 
 
 def room_proto_to_orm(
-    proto_obj: models_pb2.Room, session: sa_orm.Session
+    proto_obj: models_pb2.Room, session: eorm.Session
 ) -> Ok[eorm.Room] | orm.InvalidORMIdentifierError | InvalidArgumentError:
     orm_obj = eorm.Room(name=proto_obj.name)
     return add_orm_org_mixin_to_session(orm_obj, proto_obj, eorm.RoomID, session)
 
 
 def completion_model_proto_to_orm(
-    proto_obj: models_pb2.CompletionModel, session: sa_orm.Session
+    proto_obj: models_pb2.CompletionModel, session: eorm.Session
 ) -> Ok[eorm.CompletionModel] | InvalidArgumentError:
     orm_obj = eorm.CompletionModel(
         name=proto_obj.name,
@@ -538,7 +538,7 @@ def completion_model_proto_to_orm(
 
 def source_delete_orms(
     orm_ids: Sequence[eorm.SourceID],
-    session: sa_orm.Session,
+    session: eorm.Session,
 ) -> Ok[None] | InvalidArgumentError:
     feat_view_refs = list(
         session.scalars(
@@ -557,7 +557,7 @@ def source_delete_orms(
 
 
 def pipeline_delete_orms(
-    ids: Sequence[eorm.PipelineID], session: sa_orm.Session
+    ids: Sequence[eorm.PipelineID], session: eorm.Session
 ) -> Ok[None] | InvalidArgumentError:
     source_ids = [
         val[0]

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_resource.py

@@ -135,7 +135,7 @@ class Resource(StandardModel[ResourceID, models_pb2.Resource, eorm.Resource]):
         client: system.Client | None = None,
         ids: Iterable[ResourceID] | None = None,
         is_terminal: bool | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
         url: str | None = None,
     ) -> Ok[list[Resource]] | NotFoundError | InvalidArgumentError:
         """List resources."""
@@ -192,7 +192,7 @@ class Resource(StandardModel[ResourceID, models_pb2.Resource, eorm.Resource]):
         cls,
         resource_id: ResourceID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[Resource] | NotFoundError:
         client = client or Defaults.get_default_client()
         return cls.load_proto_for(resource_id, client, session).map(

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_source.py

@@ -83,7 +83,7 @@ class Source(StandardModel[SourceID, models_pb2.Source, eorm.Source]):
         cls,
         source_id: SourceID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[Self] | NotFoundError:
         client = client or Defaults.get_default_client()
         return cls.load_proto_for(source_id, client, session).map(
@@ -212,7 +212,7 @@ class Source(StandardModel[SourceID, models_pb2.Source, eorm.Source]):
         created_before: datetime.datetime | None = None,
         client: system.Client | None = None,
         ids: Iterable[SourceID] | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[Source]] | NotFoundError | InvalidArgumentError:
         """List sources that exist in storage."""
         client = client or Defaults.get_default_client()

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/emodel/_space.py

@@ -37,13 +37,15 @@ _DEFAULT_CONCAT_SEPARATOR = " "
 
 
 embedding_model_proto_to_name: Final[dict[embedding_models_pb2.Model, str]] = {
-    embedding_models_pb2.MODEL_CUSTOM: "random",
+    embedding_models_pb2.MODEL_CUSTOM: system.RandomTextEmbedder.model_name(),
     embedding_models_pb2.MODEL_SENTENCE_TRANSFORMER: "text-embedding-004",
     embedding_models_pb2.MODEL_GCP_TEXT_EMBEDDING_004: "text-embedding-004",
     embedding_models_pb2.MODEL_OPENAI_TEXT_EMBEDDING_3_SMALL: "text-embedding-3-small",
     embedding_models_pb2.MODEL_OPENAI_TEXT_EMBEDDING_3_LARGE: "text-embedding-3-large",
-    embedding_models_pb2.MODEL_IDENTITY: "identity",
+    embedding_models_pb2.MODEL_IDENTITY: system.IdentityTextEmbedder.model_name(),
     embedding_models_pb2.MODEL_UNSPECIFIED: "",
+    embedding_models_pb2.MODEL_CLIP: system.ClipText.model_name(),
+    embedding_models_pb2.MODEL_SIGLIP2: system.SigLIP2Text.model_name(),
 }
 name_to_proto_embedding_model = {
     name: model for model, name in embedding_model_proto_to_name.items()
@@ -265,7 +267,7 @@ class Space(StandardModel[SpaceID, models_pb2.Space, eorm.Space]):
         cls,
         space_id: SpaceID,
         client: system.Client | None = None,
-        session: sa_orm.Session | None = None,
+        session: eorm.Session | None = None,
     ) -> Ok[SpecificSpace] | NotFoundError:
         client = client or Defaults.get_default_client()
         return cls.load_proto_for(space_id, client, session).map(
@@ -284,7 +286,7 @@ class Space(StandardModel[SpaceID, models_pb2.Space, eorm.Space]):
         created_before: datetime.datetime | None = None,
         client: system.Client | None = None,
         ids: Iterable[eorm.SpaceID] | None = None,
-        existing_session: sa_orm.Session | None = None,
+        existing_session: eorm.Session | None = None,
     ) -> Ok[list[SpecificSpace]] | InvalidArgumentError | NotFoundError:
         client = client or Defaults.get_default_client()
 

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/orm/__init__.py

@@ -23,6 +23,7 @@ from corvic.orm.errors import (
     DeletedObjectError,
     InvalidORMIdentifierError,
     RequestedObjectsForNobodyError,
+    dbapi_error_to_result,
 )
 from corvic.orm.func import utc_now
 from corvic.orm.ids import (
@@ -278,4 +279,5 @@ __all__ = [
     "primary_key_foreign_column",
     "primary_key_identity_column",
     "primary_key_uuid_column",
+    "dbapi_error_to_result",
 ]

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/orm/_soft_delete.py

@@ -1,6 +1,7 @@
 """Mixin models for corvic orm tables."""
 
 from datetime import UTC, datetime
+from types import TracebackType
 from typing import Any, LiteralString
 
 import sqlalchemy as sa
@@ -9,9 +10,7 @@ from sqlalchemy import orm as sa_orm
 from sqlalchemy.ext import hybrid
 from sqlalchemy.ext.hybrid import hybrid_property
 
-from corvic.orm.errors import (
-    DeletedObjectError,
-)
+from corvic.orm.errors import DeletedObjectError, dbapi_error_to_result
 
 
 class BadDeleteError(DeletedObjectError):
@@ -155,6 +154,16 @@ class Session(sa_orm.Session):
 
     _soft_deleted: dict[sa_orm.InstanceState[Any], Any] | None = None
 
+    def __exit__(
+        self,
+        type_: type[BaseException] | None,
+        value: BaseException | None,
+        traceback: TracebackType | None,
+    ):
+        super().__exit__(type_, value, traceback)
+        if isinstance(value, sa.exc.DBAPIError):
+            raise dbapi_error_to_result(value) from value
+
     def _track_soft_deleted(self, instance: object):
         if self._soft_deleted is None:
             self._soft_deleted = {}

corvic_engine-0.3.0rc87/python/corvic/orm/errors.py

@@ -0,0 +1,54 @@
+"""Errors specific to communicating with the database."""
+
+import sqlalchemy as sa
+
+from corvic import result
+
+
+class InvalidORMIdentifierError(result.InvalidArgumentError):
+    """Raised when an identifier can't be translated to its orm equivalent."""
+
+
+class RequestedObjectsForNobodyError(result.Error):
+    """Raised when attempts are made to access database objects as the nobody org."""
+
+
+class DeletedObjectError(result.Error):
+    """DeletedObjectError result Error.
+
+    Raised on invalid operations to objects which are soft deleted.
+    """
+
+
+def dbapi_error_to_result(err: sa.exc.DBAPIError):
+    # based on https://docs.sqlalchemy.org/en/20/errors.html
+    match err:
+        case sa.exc.NotSupportedError():
+            # raised in the unexpected case that we're doing something that the
+            # database just doesn't support
+            raise result.InternalError.from_(err) from err
+        case (
+            sa.exc.OperationalError()
+            | sa.exc.InterfaceError()
+            | sa.exc.InternalError()
+        ):
+            # These are commonly things that are outside of our control that might
+            # succeed on retry, e.g., connections being dropped
+            return result.UnavailableError.from_(err)
+        case sa.exc.IntegrityError:
+            return result.InvalidArgumentError.from_(err)
+        case sa.exc.ProgrammingError():
+            if "could not serialize" in str(err):
+                return result.UnavailableError.from_(err)
+        case _:
+            pass
+
+    try:
+        import psycopg.errors
+
+        if isinstance(err.orig, psycopg.errors.SerializationFailure):
+            return result.UnavailableError.from_(err)
+    except ModuleNotFoundError:
+        pass
+
+    return result.InvalidArgumentError.from_(err)

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/system/_embedder.py

@@ -39,6 +39,9 @@ class EmbedTextResult:
 class TextEmbedder(Protocol):
     """Use a model to embed text."""
 
+    @classmethod
+    def model_name(cls) -> str: ...
+
     async def aembed(
         self,
         context: EmbedTextContext,
@@ -86,8 +89,12 @@ class LoadedModels:
 class HFModelText(TextEmbedder):
     """Generic text/image embedder from hugging face models."""
 
+    @classmethod
     @abc.abstractmethod
-    def _load_models(self) -> LoadedModels: ...
+    def model_revision(cls) -> str: ...
+
+    @abc.abstractmethod
+    def load_models(self) -> LoadedModels: ...
 
     def embed(
         self, context: EmbedTextContext
@@ -98,7 +105,7 @@ class HFModelText(TextEmbedder):
             case 32:
                 coord_dtype = pl.Float32()
 
-        models = self._load_models()
+        models = self.load_models()
         model = models.model
         processor = models.processor
         model.eval()  # type: ignore[reportAttributeAccess]
@@ -152,7 +159,15 @@ class ClipText(HFModelText):
     overcoming several major challenges in computer vision.
     """
 
-    def _load_models(self):
+    @classmethod
+    def model_name(cls) -> str:
+        return "openai/clip-vit-base-patch32"
+
+    @classmethod
+    def model_revision(cls) -> str:
+        return "5812e510083bb2d23fa43778a39ac065d205ed4d"
+
+    def load_models(self):
         from transformers.models.clip import (
             CLIPModel,
             CLIPProcessor,
@@ -161,15 +176,15 @@ class ClipText(HFModelText):
         model = cast(
             AutoModel,
             CLIPModel.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
-                pretrained_model_name_or_path="openai/clip-vit-base-patch32",
-                revision="5812e510083bb2d23fa43778a39ac065d205ed4d",
+                pretrained_model_name_or_path=self.model_name(),
+                revision=self.model_revision(),
             ),
         )
         processor = cast(
             AutoProcessor,
             CLIPProcessor.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
-                pretrained_model_name_or_path="openai/clip-vit-base-patch32",
-                revision="5812e510083bb2d23fa43778a39ac065d205ed4d",
+                pretrained_model_name_or_path=self.model_name(),
+                revision=self.model_revision(),
                 use_fast=False,
             ),
         )
@@ -179,23 +194,31 @@ class ClipText(HFModelText):
 class SigLIP2Text(HFModelText):
     """SigLIP2 text/image embedder."""
 
-    def _load_models(self):
+    @classmethod
+    def model_name(cls) -> str:
+        return "google/siglip2-base-patch16-512"
+
+    @classmethod
+    def model_revision(cls) -> str:
+        return "a89f5c5093f902bf39d3cd4d81d2c09867f0724b"
+
+    def load_models(self):
         from transformers.models.auto.modeling_auto import AutoModel
         from transformers.models.auto.processing_auto import AutoProcessor
 
         model = cast(
             AutoModel,
             AutoModel.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
-                pretrained_model_name_or_path="google/siglip2-base-patch16-512",
-                revision="a89f5c5093f902bf39d3cd4d81d2c09867f0724b",
+                pretrained_model_name_or_path=self.model_name(),
+                revision=self.model_revision(),
                 device_map="auto",
             ),
         )
         processor = cast(
             AutoProcessor,
             AutoProcessor.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
-                pretrained_model_name_or_path="google/siglip2-base-patch16-512",
-                revision="a89f5c5093f902bf39d3cd4d81d2c09867f0724b",
+                pretrained_model_name_or_path=self.model_name(),
+                revision=self.model_revision(),
                 use_fast=True,
             ),
         )

{corvic_engine-0.3.0rc85 → corvic_engine-0.3.0rc87}/python/corvic/system/_text_embedder.py

@@ -18,6 +18,10 @@ class RandomTextEmbedder(TextEmbedder):
     Useful for testing.
     """
 
+    @classmethod
+    def model_name(cls) -> str:
+        return "random"
+
     def embed(
         self, context: EmbedTextContext
     ) -> Ok[EmbedTextResult] | InvalidArgumentError | InternalError:
@@ -62,6 +66,10 @@ class IdentityTextEmbedder(TextEmbedder):
     - The resulting list is truncated or padded to match the expected vector length.
     """
 
+    @classmethod
+    def model_name(cls) -> str:
+        return "identity"
+
     def _text_to_embedding(
         self, text: str, vector_length: int, *, normalization: bool = False
     ) -> list[float]: