Skip to content

Build and run agents you can see, understand and trust.

feat(embedding): let a per-call output_dimensionality reach the gemini config (#3404)

lihongyuan998a2aef4f1787
diff --git a/src/agentscope/embedding/_gemini/_model.py b/src/agentscope/embedding/_gemini/_model.py
index 089aa9aef..53ada87b1 100644
--- a/src/agentscope/embedding/_gemini/_model.py
+++ b/src/agentscope/embedding/_gemini/_model.py
@@ -328,18 +328,15 @@ class GeminiEmbeddingModel(EmbeddingModelBase[str | TextBlock | DataBlock]):
texts.append(item)
config_kwargs = {
+ "output_dimensionality": self.dimensions,
**self.parameters.model_dump(exclude_none=True),
**kwargs,
}
- config = types.EmbedContentConfig(
- output_dimensionality=self.dimensions,
- **config_kwargs,
- )
+ config = types.EmbedContentConfig(**config_kwargs)
cache_key = {
"model": self.model,
"contents": texts,
- "output_dimensionality": self.dimensions,
**config_kwargs,
}
@@ -421,10 +418,11 @@ class GeminiEmbeddingModel(EmbeddingModelBase[str | TextBlock | DataBlock]):
f"Invalid input: {item!r}. Expected str or DataBlock.",
)
- config = types.EmbedContentConfig(
- output_dimensionality=self.dimensions,
+ config_kwargs = {
+ "output_dimensionality": self.dimensions,
**kwargs,
- )
+ }
+ config = types.EmbedContentConfig(**config_kwargs)
start_time = datetime.now()
response = await self.client.aio.models.embed_content(
diff --git a/tests/embedding_gemini_test.py b/tests/embedding_gemini_test.py
index be4df866b..7d842044a 100644
--- a/tests/embedding_gemini_test.py
+++ b/tests/embedding_gemini_test.py
@@ -159,11 +159,12 @@ class GeminiTextCallTest(IsolatedAsyncioTestCase):
"""Configured text options reach the SDK; call options win."""
model = self._make_configured_model("RETRIEVAL_QUERY")
sdk = model.client.aio.models.embed_content
- for overrides, task_type in (
- ({}, "RETRIEVAL_QUERY"),
- ({"task_type": "RETRIEVAL_DOCUMENT"}, "RETRIEVAL_DOCUMENT"),
+ for overrides, task_type, dimensions in (
+ ({}, "RETRIEVAL_QUERY", 2),
+ ({"task_type": "RETRIEVAL_DOCUMENT"}, "RETRIEVAL_DOCUMENT", 2),
+ ({"output_dimensionality": 4}, "RETRIEVAL_QUERY", 4),
):
- with self.subTest(task_type=task_type):
+ with self.subTest(task_type=task_type, dimensions=dimensions):
await model(["hello"], **overrides)
request = sdk.call_args.kwargs
self.assertEqual(
@@ -178,7 +179,7 @@ class GeminiTextCallTest(IsolatedAsyncioTestCase):
"contents": ["hello"],
"config": {
"task_type": task_type,
- "output_dimensionality": 2,
+ "output_dimensionality": dimensions,
},
},
)
@@ -186,19 +187,37 @@ class GeminiTextCallTest(IsolatedAsyncioTestCase):
model.parameters.model_dump(),
{"task_type": "RETRIEVAL_QUERY"},
)
- self.assertEqual(sdk.await_count, 2)
+ self.assertEqual(sdk.await_count, 3)
async def test_file_cache_uses_effective_parameters(self) -> None:
- """Distinct task configurations cannot reuse each other's vectors."""
+ """Cache by effective task and dimensions, reusing matching vectors."""
with TemporaryDirectory() as directory:
cache = FileEmbeddingCache(cache_dir=directory)
query = self._make_configured_model("RETRIEVAL_QUERY", cache)
document = self._make_configured_model("RETRIEVAL_DOCUMENT", cache)
+ query.client.aio.models.embed_content.side_effect = [
+ MagicMock(embeddings=[MagicMock(values=[0.1, 0.2])]),
+ MagicMock(
+ embeddings=[MagicMock(values=[0.1, 0.2, 0.3, 0.4])],
+ ),
+ ]
document.client.aio.models.embed_content.return_value = MagicMock(
embeddings=[MagicMock(values=[0.3, 0.4])],
)
cases = (
(query, {}, "api", [[0.1, 0.2]]),
+ (
+ query,
+ {"output_dimensionality": 4},
+ "api",
+ [[0.1, 0.2, 0.3, 0.4]],
+ ),
+ (
+ query,
+ {"output_dimensionality": 4},
+ "cache",
+ [[0.1, 0.2, 0.3, 0.4]],
+ ),
(document, {}, "api", [[0.3, 0.4]]),
(
query,
@@ -207,6 +226,12 @@ class GeminiTextCallTest(IsolatedAsyncioTestCase):
[[0.3, 0.4]],
),
(query, {}, "cache", [[0.1, 0.2]]),
+ (
+ query,
+ {"output_dimensionality": 2},
+ "cache",
+ [[0.1, 0.2]],
+ ),
)
for index, (model, overrides, source, embeddings) in enumerate(
cases,
@@ -228,7 +253,10 @@ class GeminiTextCallTest(IsolatedAsyncioTestCase):
"source": source,
},
)
- query.client.aio.models.embed_content.assert_awaited_once()
+ self.assertEqual(
+ query.client.aio.models.embed_content.await_count,
+ 2,
+ )
document.client.aio.models.embed_content.assert_awaited_once()
@@ -276,6 +304,34 @@ class GeminiMultimodalCallTest(IsolatedAsyncioTestCase):
},
)
+ async def test_multimodal_call_accepts_per_call_dimensions(self) -> None:
+ """Multimodal mode forwards the override into the request config."""
+ model = self._make_multimodal_model()
+ model.client = MagicMock()
+ model.client.aio.models.embed_content = AsyncMock(
+ return_value=MagicMock(
+ embeddings=[MagicMock(values=[0.1, 0.2, 0.3, 0.4])],
+ ),
+ )
+ await model(["hello"], output_dimensionality=4)
+
+ request = model.client.aio.models.embed_content.call_args.kwargs
+ self.assertEqual(
+ {
+ **request,
+ "contents": [
+ item.model_dump(exclude_none=True)
+ for item in request["contents"]
+ ],
+ "config": request["config"].model_dump(exclude_none=True),
+ },
+ {
+ "model": "gemini-embedding-2",
+ "contents": [{"parts": [{"text": "hello"}]}],
+ "config": {"output_dimensionality": 4},
+ },
+ )
+
async def test_multimodal_batching_by_image_limit(self) -> None:
"""8 images with max_images=6 produces 2 batches (6+2)."""
model = self._make_multimodal_model()