mirror of
https://github.com/hwchase17/langchain.git
synced 2025-09-29 15:28:54 +00:00
844955d6e1
Assigning missed defaults in various classes. Most clients were being
assigned during the `model_validator(mode="before")` step, so this
change should amount to a no-op in those cases.
---
This PR was autogenerated using gritql
```shell
grit apply 'class_definition(name=$C, $body, superclasses=$S) where {
$C <: ! "Config", // Does not work in this scope, but works after class_definition
$body <: block($statements),
$statements <: some bubble assignment(left=$x, right=$y, type=$t) as $A where {
or {
$y <: `Field($z)`,
$x <: "model_config"
}
},
// And has either Any or Optional fields without a default
$statements <: some bubble assignment(left=$x, right=$y, type=$t) as $A where {
$t <: or {
r"Optional.*",
r"Any",
r"Union[None, .*]",
r"Union[.*, None, .*]",
r"Union[.*, None]",
},
$y <: ., // Match empty node
$t => `$t = None`,
},
}
' --language python .
```
65 lines
2.1 KiB
Python
65 lines
2.1 KiB
Python
from typing import Any, Dict, List, Tuple
|
|
|
|
from pydantic import BaseModel, ConfigDict, Field
|
|
|
|
from langchain_community.cross_encoders.base import BaseCrossEncoder
|
|
|
|
DEFAULT_MODEL_NAME = "BAAI/bge-reranker-base"
|
|
|
|
|
|
class HuggingFaceCrossEncoder(BaseModel, BaseCrossEncoder):
|
|
"""HuggingFace cross encoder models.
|
|
|
|
Example:
|
|
.. code-block:: python
|
|
|
|
from langchain_community.cross_encoders import HuggingFaceCrossEncoder
|
|
|
|
model_name = "BAAI/bge-reranker-base"
|
|
model_kwargs = {'device': 'cpu'}
|
|
hf = HuggingFaceCrossEncoder(
|
|
model_name=model_name,
|
|
model_kwargs=model_kwargs
|
|
)
|
|
"""
|
|
|
|
client: Any = None #: :meta private:
|
|
model_name: str = DEFAULT_MODEL_NAME
|
|
"""Model name to use."""
|
|
model_kwargs: Dict[str, Any] = Field(default_factory=dict)
|
|
"""Keyword arguments to pass to the model."""
|
|
|
|
def __init__(self, **kwargs: Any):
|
|
"""Initialize the sentence_transformer."""
|
|
super().__init__(**kwargs)
|
|
try:
|
|
import sentence_transformers
|
|
|
|
except ImportError as exc:
|
|
raise ImportError(
|
|
"Could not import sentence_transformers python package. "
|
|
"Please install it with `pip install sentence-transformers`."
|
|
) from exc
|
|
|
|
self.client = sentence_transformers.CrossEncoder(
|
|
self.model_name, **self.model_kwargs
|
|
)
|
|
|
|
model_config = ConfigDict(extra="forbid", protected_namespaces=())
|
|
|
|
def score(self, text_pairs: List[Tuple[str, str]]) -> List[float]:
|
|
"""Compute similarity scores using a HuggingFace transformer model.
|
|
|
|
Args:
|
|
text_pairs: The list of text text_pairs to score the similarity.
|
|
|
|
Returns:
|
|
List of scores, one for each pair.
|
|
"""
|
|
scores = self.client.predict(text_pairs)
|
|
# Some models e.g bert-multilingual-passage-reranking-msmarco
|
|
# gives two score not_relevant and relevant as compare with the query.
|
|
if len(scores.shape) > 1: # we are going to get the relevant scores
|
|
scores = map(lambda x: x[1], scores)
|
|
return scores
|