wehub-resource-sync
114 lines
3.7 KiB
Python
114 lines
3.7 KiB
Python
from typing import List, Optional, Dict
|
|
|
|
from deepeval.config.settings import get_settings
|
|
from deepeval.utils import require_dependency
|
|
from deepeval.models import DeepEvalBaseEmbeddingModel
|
|
from deepeval.models.utils import (
|
|
normalize_kwargs_and_extract_aliases,
|
|
)
|
|
from deepeval.models.retry_policy import (
|
|
create_retry_decorator,
|
|
)
|
|
from deepeval.constants import ProviderSlug as PS
|
|
from deepeval.utils import require_param
|
|
|
|
retry_ollama = create_retry_decorator(PS.OLLAMA)
|
|
|
|
_ALIAS_MAP = {"base_url": ["host"]}
|
|
|
|
|
|
class OllamaEmbeddingModel(DeepEvalBaseEmbeddingModel):
|
|
def __init__(
|
|
self,
|
|
model: Optional[str] = None,
|
|
base_url: Optional[str] = None,
|
|
generation_kwargs: Optional[Dict] = None,
|
|
**kwargs,
|
|
):
|
|
normalized_kwargs, alias_values = normalize_kwargs_and_extract_aliases(
|
|
"OllamaEmbeddingModel",
|
|
kwargs,
|
|
_ALIAS_MAP,
|
|
)
|
|
|
|
# re-map depricated keywords to re-named positional args
|
|
if base_url is None and "base_url" in alias_values:
|
|
base_url = alias_values["base_url"]
|
|
|
|
settings = get_settings()
|
|
|
|
if base_url is not None:
|
|
self.base_url = str(base_url).rstrip("/")
|
|
elif settings.LOCAL_EMBEDDING_BASE_URL is not None:
|
|
self.base_url = str(settings.LOCAL_EMBEDDING_BASE_URL).rstrip("/")
|
|
else:
|
|
self.base_url = "http://localhost:11434"
|
|
|
|
model = model or settings.LOCAL_EMBEDDING_MODEL_NAME
|
|
|
|
# validation
|
|
model = require_param(
|
|
model,
|
|
provider_label="OllamaEmbeddingModel",
|
|
env_var_name="LOCAL_EMBEDDING_MODEL_NAME",
|
|
param_hint="model",
|
|
)
|
|
|
|
# Keep sanitized kwargs for client call to strip legacy keys
|
|
self.kwargs = normalized_kwargs
|
|
self.generation_kwargs = generation_kwargs or {}
|
|
super().__init__(model)
|
|
|
|
@retry_ollama
|
|
def embed_text(self, text: str) -> List[float]:
|
|
embedding_model = self.load_model()
|
|
response = embedding_model.embed(
|
|
model=self.name, input=text, **self.generation_kwargs
|
|
)
|
|
return response["embeddings"][0]
|
|
|
|
@retry_ollama
|
|
def embed_texts(self, texts: List[str]) -> List[List[float]]:
|
|
embedding_model = self.load_model()
|
|
response = embedding_model.embed(
|
|
model=self.name, input=texts, **self.generation_kwargs
|
|
)
|
|
return response["embeddings"]
|
|
|
|
@retry_ollama
|
|
async def a_embed_text(self, text: str) -> List[float]:
|
|
embedding_model = self.load_model(async_mode=True)
|
|
response = await embedding_model.embed(
|
|
model=self.name, input=text, **self.generation_kwargs
|
|
)
|
|
return response["embeddings"][0]
|
|
|
|
@retry_ollama
|
|
async def a_embed_texts(self, texts: List[str]) -> List[List[float]]:
|
|
embedding_model = self.load_model(async_mode=True)
|
|
response = await embedding_model.embed(
|
|
model=self.name, input=texts, **self.generation_kwargs
|
|
)
|
|
return response["embeddings"]
|
|
|
|
###############################################
|
|
# Model
|
|
###############################################
|
|
|
|
def load_model(self, async_mode: bool = False):
|
|
ollama = require_dependency(
|
|
"ollama",
|
|
provider_label="OllamaEmbeddingModel",
|
|
install_hint="Install it with `pip install ollama`.",
|
|
)
|
|
|
|
if not async_mode:
|
|
return self._build_client(ollama.Client)
|
|
return self._build_client(ollama.AsyncClient)
|
|
|
|
def _build_client(self, cls):
|
|
return cls(host=self.base_url, **self.kwargs)
|
|
|
|
def get_model_name(self):
|
|
return f"{self.name} (Ollama)"
|