support new dense vector quantization in 8.16

Author: miguelgrinberg

elasticsearch_dsl/field.py

@@ -389,13 +389,23 @@ def _deserialize(self, data: Any) -> float:
         return float(data)
 
 
-class DenseVector(Float):
+class DenseVector(Field):
     name = "dense_vector"
+    _coerce = True
 
     def __init__(self, **kwargs: Any):
-        kwargs["multi"] = True
+        self._element_type = kwargs.get("element_type", "float")
+        if self._element_type in ["float", "byte"]:
+            kwargs["multi"] = True
         super().__init__(**kwargs)
 
+    def _deserialize(self, data: Any) -> float:
+        if self._element_type == "float":
+            return float(data)
+        elif self._element_type == "byte":
+            return int(data)
+        return data
+
 
 class SparseVector(Field):
     name = "sparse_vector"

tests/test_integration/_async/test_document.py

@@ -37,6 +37,7 @@
     Binary,
     Boolean,
     Date,
+    DenseVector,
     Double,
     InnerDoc,
     Ip,
@@ -795,3 +796,28 @@ async def gen3() -> AsyncIterator[Union[Doc, Dict[str, Any]]]:
         "age": 45,
         "languages": ["es"],
     }
+
+
+@pytest.mark.asyncio
+async def test_dense_vector_quantization(async_client: AsyncElasticsearch) -> None:
+    class Doc(AsyncDocument):
+        float_vector: List[float] = mapped_field(DenseVector())
+        byte_vector: List[int] = mapped_field(DenseVector(element_type="byte"))
+        bit_vector: str = mapped_field(DenseVector(element_type="bit"))
+
+        class Index:
+            name = "vectors"
+
+    await Doc._index.delete(ignore_unavailable=True)
+    await Doc.init()
+
+    doc = Doc(
+        float_vector=[1.0, 1.2, 2.3], byte_vector=[12, 23, 34, 45], bit_vector="12abf0"
+    )
+    await doc.save(refresh=True)
+
+    docs = await Doc.search().execute()
+    assert len(docs) == 1
+    assert docs[0].float_vector == doc.float_vector
+    assert docs[0].byte_vector == doc.byte_vector
+    assert docs[0].bit_vector == doc.bit_vector

tests/test_integration/_sync/test_document.py

@@ -35,6 +35,7 @@
     Binary,
     Boolean,
     Date,
+    DenseVector,
     Document,
     Double,
     InnerDoc,
@@ -789,3 +790,28 @@ def gen3() -> Iterator[Union[Doc, Dict[str, Any]]]:
         "age": 45,
         "languages": ["es"],
     }
+
+
+@pytest.mark.sync
+def test_dense_vector_quantization(client: Elasticsearch) -> None:
+    class Doc(Document):
+        float_vector: List[float] = mapped_field(DenseVector())
+        byte_vector: List[int] = mapped_field(DenseVector(element_type="byte"))
+        bit_vector: str = mapped_field(DenseVector(element_type="bit"))
+
+        class Index:
+            name = "vectors"
+
+    Doc._index.delete(ignore_unavailable=True)
+    Doc.init()
+
+    doc = Doc(
+        float_vector=[1.0, 1.2, 2.3], byte_vector=[12, 23, 34, 45], bit_vector="12abf0"
+    )
+    doc.save(refresh=True)
+
+    docs = Doc.search().execute()
+    assert len(docs) == 1
+    assert docs[0].float_vector == doc.float_vector
+    assert docs[0].byte_vector == doc.byte_vector
+    assert docs[0].bit_vector == doc.bit_vector