aryn-ai · bsowell · May 14, 2025 · May 11, 2025 · May 11, 2025 · May 13, 2025
diff --git a/examples/simple_weaviate.py b/examples/simple_weaviate.py
@@ -1,7 +1,8 @@
 import sys
 
 from weaviate.classes.config import ReferenceProperty
-from weaviate.client import AdditionalConfig, ConnectionParams
+from weaviate.config import AdditionalConfig
+from weaviate.connect.base import ConnectionParams
 from weaviate.collections.classes.config import Configure
 from weaviate.config import Timeout
 

diff --git a/lib/poetry-lock/poetry.lock b/lib/poetry-lock/poetry.lock
diff --git a/lib/poetry-lock/pyproject.toml b/lib/poetry-lock/pyproject.toml
@@ -9,7 +9,6 @@ repository = "https://github.com/aryn-ai/sycamore.git"
 [tool.poetry.dependencies]
 python = ">=3.9.2,<3.13"
 # required older version
-protobuf = "4.25.3"   # rps
 fsspec = "2024.2.0"   # sycamore (via datasets)
 # Make sure tests are run in the same order.
 # the following will fail which prevents upgrading pytest to 8.1.1
@@ -18,7 +17,7 @@ pytest = "7.4.0"
 guidance = "0.1.14" # They removed AzureOpenAIChat in https://github.com/guidance-ai/guidance/commit/a75896a851edf8b3cf15534af4fed8d15f277af1
 ray = ">=2.41.0" # Latest version supported on linux/aarch64 https://github.com/ray-project/ray/issues/46104
 pandas = "2.1.1" # Hardcoded into sycamore pyproject.toml. Unclear why.
-weaviate-client = "4.6.4" # Failes typechecking with 4.7.0
+weaviate-client = "4.14.1" # Failes typechecking with 4.7.0
 
 # Required to continue supporting python 3.9; it may be possible to make these a bit more
 # current, but latest version no longer supports 3.9

diff --git a/lib/sycamore/poetry.lock b/lib/sycamore/poetry.lock
diff --git a/lib/sycamore/pyproject.toml b/lib/sycamore/pyproject.toml
@@ -68,9 +68,9 @@ elasticsearch = { version = "8.14.0", optional = true}
 duckdb = { version = "^1.1.1", optional = true }
 neo4j = { version = "^5.21.0", optional = true }
 opensearch-py = { version = "^2.3.1", optional = true }
-pinecone-client = { version = "^4.1.0", extras = ["grpc"], optional = true }
+pinecone = { version = "^6.0.2", extras = ["grpc"], optional = true }
 pinecone-text = { version = "^0.9.0", optional = true }
-weaviate-client = { version = ">=4.6.1,<4.7", optional = true } # typechecking fails with 4.7
+weaviate-client = { version = ">=4.14.1", optional = true } # typechecking fails with 4.7
 qdrant-client = { version = "^1.11.2", optional = true }
 
 # Local inference dependencies
@@ -92,7 +92,7 @@ nltk = { version = "^3.9.0", optional = true }
 
 # LLM Dependencies
 anthropic = { version = "^0.42.0", optional = true }
-google-genai = {version = "^1.0", optional = true}
+google-genai = {version = "^1.14.0", optional = true}
 tiktoken = "^0.8.0"
 
 [tool.poetry.group.test.dependencies]
@@ -129,7 +129,7 @@ duckdb = ["duckdb"]
 elasticsearch = ["elasticsearch"]
 neo4j = ["neo4j"]
 opensearch = ["opensearch-py"]
-pinecone = ["pinecone-client", "pinecone-text"]
+pinecone = ["pinecone", "pinecone-text"]
 weaviate = ["weaviate-client"]
 qdrant = ["qdrant-client"]
 

diff --git a/lib/sycamore/sycamore/connectors/pinecone/pinecone_reader.py b/lib/sycamore/sycamore/connectors/pinecone/pinecone_reader.py
@@ -44,7 +44,7 @@ def read_records(self, query_params: BaseDBReader.QueryParams) -> "PineconeReade
             for pids in index.list(namespace=query_params.namespace):
                 ids.extend(pids)
             results = PineconeReaderQueryResponse(
-                list(dict(index.fetch(ids=ids, namespace=query_params.namespace)["vectors"]).values())
+                list(dict(index.fetch(ids=ids, namespace=query_params.namespace).vectors).values())
             )
         return results
 

diff --git a/lib/sycamore/sycamore/connectors/weaviate/weaviate_reader.py b/lib/sycamore/sycamore/connectors/weaviate/weaviate_reader.py
@@ -10,12 +10,10 @@
 
 if typing.TYPE_CHECKING:
     from weaviate import WeaviateClient
-    from weaviate.client import (
-        AdditionalConfig,
-        AuthCredentials,
-        ConnectionParams,
-        EmbeddedOptions,
-    )
+    from weaviate.auth import AuthCredentials
+    from weaviate.connect.base import ConnectionParams
+    from weaviate.config import AdditionalConfig
+    from weaviate.embedded import EmbeddedOptions
 
 
 @dataclass
@@ -66,7 +64,7 @@ def read_records(self, query_params: BaseDBReader.QueryParams) -> "WeaviateReade
             else:
                 collection = list(
                     self._client.collections.get(query_params.collection_name).iterator(include_vector=True)
-                )
+                )  # type: ignore
             results = WeaviateReaderQueryResponse(collection=collection)
             return results
 

diff --git a/lib/sycamore/sycamore/connectors/weaviate/weaviate_writer.py b/lib/sycamore/sycamore/connectors/weaviate/weaviate_writer.py
@@ -12,16 +12,14 @@
 
 if typing.TYPE_CHECKING:
     from weaviate import WeaviateClient
-    from weaviate.client import (
-        AdditionalConfig,
-        AuthCredentials,
-        ConnectionParams,
-        EmbeddedOptions,
-    )
+    from weaviate.auth import AuthCredentials
     from weaviate.collections.classes.config import (
         _CollectionConfigCreate,
         CollectionConfig,
     )
+    from weaviate.connect.base import ConnectionParams
+    from weaviate.config import AdditionalConfig
+    from weaviate.embedded import EmbeddedOptions
 
 
 # This is a convenience so that you can use the alias CollectionConfigCreate
@@ -127,7 +125,7 @@ def write_many_records(self, records: list[BaseDBWriter.Record], target_params:
                         batch.add_object(properties=r.properties, uuid=r.uuid)
 
     def create_target_idempotent(self, target_params: BaseDBWriter.TargetParams):
-        from weaviate.client import UnexpectedStatusCodeError
+        from weaviate.exceptions import UnexpectedStatusCodeError
         from weaviate.collections.classes.config import CollectionConfig
 
         assert isinstance(target_params, WeaviateWriterTargetParams)

diff --git a/lib/sycamore/sycamore/llms/config.py b/lib/sycamore/sycamore/llms/config.py
@@ -54,6 +54,7 @@ class GeminiModels(Enum):
     """Represents available Gemini models. More info: https://googleapis.github.io/python-genai/"""
 
     # Note that the models available on a given Gemini account may vary.
+    GEMINI_2_5_FLASH_PREVIEW = GeminiModel(name="gemini-2.5-flash-preview-04-17", is_chat=True)
     GEMINI_2_FLASH = GeminiModel(name="gemini-2.0-flash", is_chat=True)
     GEMINI_2_FLASH_LITE = GeminiModel(name="gemini-2.0-flash-lite", is_chat=True)
     GEMINI_2_FLASH_THINKING = GeminiModel(name="gemini-2.0-flash-thinking-exp", is_chat=True)

diff --git a/lib/sycamore/sycamore/llms/gemini.py b/lib/sycamore/sycamore/llms/gemini.py
@@ -85,9 +85,12 @@ def get_generate_kwargs(self, prompt: RenderedPrompt, llm_kwargs: Optional[dict]
                     buffered = io.BytesIO()
                     image.save(buffered, format="PNG")
                     image_bytes = buffered.getvalue()
+                    assert content.parts is not None  # mypy
                     content.parts.append(types.Part.from_bytes(data=image_bytes, mime_type="image/png"))
             content_list.append(content)
         kwargs["config"] = None
+        if thinking_budget := config.pop("thinking_budget", None):
+            config["thinking_config"] = types.ThinkingConfig(thinking_budget=thinking_budget)
         if config:
             kwargs["config"] = types.GenerateContentConfig(**config)
         kwargs["content"] = content_list

diff --git a/lib/sycamore/sycamore/tests/integration/connectors/pinecone/test_pinecone_read.py b/lib/sycamore/sycamore/tests/integration/connectors/pinecone/test_pinecone_read.py
@@ -70,7 +70,7 @@ def wait_for_write_completion(client: PineconeGRPC, index_name: str, namespace:
     index = client.Index(index_name)
     while time.time() < deadline:
         try:
-            desc = dict(index.fetch(ids=[doc_id], namespace=namespace)["vectors"]).items()
+            desc = dict(index.fetch(ids=[doc_id], namespace=namespace).vectors).items()
             if len(desc) > 0:
                 return
         except PineconeException:

diff --git a/lib/sycamore/sycamore/tests/integration/connectors/weaviate/test_pdf_to_weaviate.py b/lib/sycamore/sycamore/tests/integration/connectors/weaviate/test_pdf_to_weaviate.py
@@ -2,7 +2,7 @@
 
 import weaviate
 from weaviate.classes.config import Property, ReferenceProperty
-from weaviate.client import ConnectionParams
+from weaviate.connect.base import ConnectionParams
 from weaviate.collections.classes.config import Configure, DataType
 
 import sycamore

diff --git a/lib/sycamore/sycamore/tests/integration/connectors/weaviate/test_weaviate_read.py b/lib/sycamore/sycamore/tests/integration/connectors/weaviate/test_weaviate_read.py
@@ -4,7 +4,7 @@
 import weaviate
 from weaviate.classes.config import Property, ReferenceProperty
 from weaviate.classes.query import Filter
-from weaviate.client import ConnectionParams
+from weaviate.connect.base import ConnectionParams
 from weaviate.collections.classes.config import Configure, DataType
 
 import sycamore

diff --git a/lib/sycamore/sycamore/tests/resources/objects/weaviate/collection_params_b.pickle b/lib/sycamore/sycamore/tests/resources/objects/weaviate/collection_params_b.pickle
diff --git a/lib/sycamore/sycamore/tests/unit/connectors/pinecone/test_pinecone_reader.py b/lib/sycamore/sycamore/tests/unit/connectors/pinecone/test_pinecone_reader.py
@@ -8,6 +8,8 @@
     PineconeReaderQueryResponse,
 )
 
+from pinecone.data import FetchResponse, Vector
+
 
 @pytest.fixture
 def client_params():
@@ -40,9 +42,11 @@ def test_pinecone_reader_client_read_records(query_params, mock_pinecone_grpc):
     mock_index = mock.Mock()
     mock_pinecone_grpc.return_value.Index.return_value = mock_index
     mock_index.list.return_value = [["id1", "id2"]]
-    mock_index.fetch.return_value = {
-        "vectors": {"id1": {"id": "id1", "values": [0.1, 0.2]}, "id2": {"id": "id2", "values": [0.3, 0.4]}}
-    }
+    mock_index.fetch.return_value = FetchResponse(
+        namespace="test_namespace",
+        vectors={"id1": Vector(id="id1", values=[0.1, 0.2]), "id2": Vector(id="id2", values=[0.3, 0.4])},
+        usage=None,
+    )
 
     client = PineconeReaderClient(PineconeReaderClientParams(api_key="test_api_key"))
     response = client.read_records(query_params)

diff --git a/lib/sycamore/sycamore/tests/unit/connectors/weaviate/test_weaviate_writer.py b/lib/sycamore/sycamore/tests/unit/connectors/weaviate/test_weaviate_writer.py
@@ -13,7 +13,7 @@
 from sycamore.tests.config import TEST_DIR
 import weaviate
 from weaviate.classes.config import Property, ReferenceProperty
-from weaviate.client import ConnectionParams
+from weaviate.connect.base import ConnectionParams
 from weaviate.collections.classes.config import Configure, DataType
 from weaviate.exceptions import WeaviateInvalidInputError