diff --git a/poetry.lock b/poetry.lock
index d6c0340d0dddce5a9e23e48709dacafda2a18ecd..668ba3e85d77d119e41d9aded3a819827bc95dd6 100644
--- a/poetry.lock
+++ b/poetry.lock
@@ -5562,4 +5562,4 @@ vision = ["pillow", "torch", "torchvision", "transformers"]
[metadata]
lock-version = "2.0"
python-versions = ">=3.9,<3.14"
-content-hash = "e88cbb248e65005bf3265b1ef31c768baf6f979d08e829458163cb567e09ffa3"
+content-hash = "7a8c31e645e649de60985ecb0ddec1aefcdeb1cfcbe23b47185a0bd18407a900"
diff --git a/pyproject.toml b/pyproject.toml
index 5a52b7745bc089fd270c3c6a77197322604a36b3..a6b71d9cef2b29cf2583b1e7d6f0f4c077f55c42 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -9,7 +9,7 @@ license = "MIT"
[tool.poetry.dependencies]
python = ">=3.9,<3.14"
-pydantic = "^2.5.3"
+pydantic = "^2.10.2"
openai = ">=1.10.0,<2.0.0"
cohere = {version = ">=5.9.4,<6.00", optional = true}
mistralai= {version = ">=0.0.12,<0.1.0", optional = true}
diff --git a/semantic_router/encoders/base.py b/semantic_router/encoders/base.py
index 0a25f21050e2e5da1ea610fb5194324cdeef9a5f..993093c1a2132c58e769b3035ed7588aefe7a689 100644
--- a/semantic_router/encoders/base.py
+++ b/semantic_router/encoders/base.py
@@ -1,6 +1,6 @@
from typing import Any, Coroutine, List, Optional
-from pydantic.v1 import BaseModel, Field, validator
+from pydantic import BaseModel, Field, field_validator
import numpy as np
from semantic_router.schema import SparseEmbedding
@@ -14,7 +14,7 @@ class DenseEncoder(BaseModel):
class Config:
arbitrary_types_allowed = True
- @validator("score_threshold", pre=True, always=True)
+ @field_validator("score_threshold")
def set_score_threshold(cls, v):
return float(v) if v is not None else None
diff --git a/semantic_router/encoders/clip.py b/semantic_router/encoders/clip.py
index 65fbdb8f0312d5339d68ae5900542aea2c896693..d30f5e1fa8c80a54ec8131f62ca99a14ce663ddf 100644
--- a/semantic_router/encoders/clip.py
+++ b/semantic_router/encoders/clip.py
@@ -9,7 +9,6 @@ from semantic_router.encoders import DenseEncoder
class CLIPEncoder(DenseEncoder):
name: str = "openai/clip-vit-base-patch16"
type: str = "huggingface"
- score_threshold: float = 0.2
tokenizer_kwargs: Dict = {}
processor_kwargs: Dict = {}
model_kwargs: Dict = {}
@@ -21,6 +20,8 @@ class CLIPEncoder(DenseEncoder):
_Image: Any = PrivateAttr()
def __init__(self, **data):
+ if data.get("score_threshold") is None:
+ data["score_threshold"] = 0.2
super().__init__(**data)
self._tokenizer, self._processor, self._model = self._initialize_hf_model()
diff --git a/semantic_router/encoders/huggingface.py b/semantic_router/encoders/huggingface.py
index 7c7e56f77f11cf5143437c822ffda338c8f92cf0..c791e1c35a24d4f42e32e9e6291101249acfb5ff 100644
--- a/semantic_router/encoders/huggingface.py
+++ b/semantic_router/encoders/huggingface.py
@@ -34,7 +34,6 @@ from semantic_router.utils.logger import logger
class HuggingFaceEncoder(DenseEncoder):
name: str = "sentence-transformers/all-MiniLM-L6-v2"
type: str = "huggingface"
- score_threshold: float = 0.5
tokenizer_kwargs: Dict = {}
model_kwargs: Dict = {}
device: Optional[str] = None
@@ -43,6 +42,8 @@ class HuggingFaceEncoder(DenseEncoder):
_torch: Any = PrivateAttr()
def __init__(self, **data):
+ if data.get("score_threshold") is None:
+ data["score_threshold"] = 0.5
super().__init__(**data)
self._tokenizer, self._model = self._initialize_hf_model()
@@ -153,7 +154,6 @@ class HFEndpointEncoder(DenseEncoder):
name: str = "hugging_face_custom_endpoint"
huggingface_url: Optional[str] = None
huggingface_api_key: Optional[str] = None
- score_threshold: float = 0.8
def __init__(
self,
@@ -180,6 +180,8 @@ class HFEndpointEncoder(DenseEncoder):
"""
huggingface_url = huggingface_url or os.getenv("HF_API_URL")
huggingface_api_key = huggingface_api_key or os.getenv("HF_API_KEY")
+ if score_threshold is None:
+ score_threshold = 0.8
super().__init__(name=name, score_threshold=score_threshold) # type: ignore
diff --git a/semantic_router/encoders/vit.py b/semantic_router/encoders/vit.py
index 73cb058281d3c47d452bfa3d5a8e4c92246ae173..dec768e43d3eff3a8eb6a428ea4e5773a488c9ad 100644
--- a/semantic_router/encoders/vit.py
+++ b/semantic_router/encoders/vit.py
@@ -1,6 +1,6 @@
-from typing import Any, List, Optional, Dict
+from typing import Any, Dict, List, Optional
-from pydantic.v1 import PrivateAttr
+from pydantic import PrivateAttr
from semantic_router.encoders import DenseEncoder
@@ -8,7 +8,6 @@ from semantic_router.encoders import DenseEncoder
class VitEncoder(DenseEncoder):
name: str = "google/vit-base-patch16-224"
type: str = "huggingface"
- score_threshold: float = 0.5
processor_kwargs: Dict = {}
model_kwargs: Dict = {}
device: Optional[str] = None
@@ -19,6 +18,8 @@ class VitEncoder(DenseEncoder):
_Image: Any = PrivateAttr()
def __init__(self, **data):
+ if data.get("score_threshold") is None:
+ data["score_threshold"] = 0.5
super().__init__(**data)
self._processor, self._model = self._initialize_hf_model()
diff --git a/semantic_router/index/postgres.py b/semantic_router/index/postgres.py
index 6d445f2bedabeba63a24712db955d6d4757647a2..71ea32e85fb84531f85178fd74663ad8bce61bcb 100644
--- a/semantic_router/index/postgres.py
+++ b/semantic_router/index/postgres.py
@@ -103,7 +103,6 @@ class PostgresIndex(BaseIndex):
connection_string: Optional[str] = None
index_prefix: str = "semantic_router_"
index_name: str = "index"
- dimensions: int = 1536
metric: Metric = Metric.COSINE
namespace: Optional[str] = ""
conn: Optional["psycopg2.extensions.connection"] = None
@@ -115,9 +114,9 @@ class PostgresIndex(BaseIndex):
connection_string: Optional[str] = None,
index_prefix: str = "semantic_router_",
index_name: str = "index",
- dimensions: int = 1536,
metric: Metric = Metric.COSINE,
namespace: Optional[str] = "",
+ dimensions: int | None = None,
):
"""
Initializes the Postgres index with the specified parameters.
@@ -135,6 +134,8 @@ class PostgresIndex(BaseIndex):
:param namespace: An optional namespace for the index.
:type namespace: Optional[str]
"""
+ if dimensions is None:
+ dimensions = 1536
super().__init__()
# try and import psycopg2
try: