langchain/libs/community/langchain_community/embeddings/ipex_llm.py

141 lines
5.1 KiB
Python
Raw Normal View History

# This file is adapted from
# https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/embeddings/huggingface.py
from typing import Any, Dict, List, Optional
from langchain_core.embeddings import Embeddings
from langchain_core.pydantic_v1 import BaseModel, Extra, Field
DEFAULT_BGE_MODEL = "BAAI/bge-small-en-v1.5"
DEFAULT_QUERY_BGE_INSTRUCTION_EN = (
"Represent this question for searching relevant passages: "
)
DEFAULT_QUERY_BGE_INSTRUCTION_ZH = "为这个句子生成表示以用于检索相关文章:"
class IpexLLMBgeEmbeddings(BaseModel, Embeddings):
"""Wrapper around the BGE embedding model
with IPEX-LLM optimizations on Intel CPUs and GPUs.
To use, you should have the ``ipex-llm``
and ``sentence_transformers`` package installed. Refer to
`here <https://python.langchain.com/v0.1/docs/integrations/text_embedding/ipex_llm/>`_
for installation on Intel CPU.
Example on Intel CPU:
.. code-block:: python
from langchain_community.embeddings import IpexLLMBgeEmbeddings
embedding_model = IpexLLMBgeEmbeddings(
model_name="BAAI/bge-large-en-v1.5",
model_kwargs={},
encode_kwargs={"normalize_embeddings": True},
)
Refer to
`here <https://python.langchain.com/v0.1/docs/integrations/text_embedding/ipex_llm_gpu/>`_
for installation on Intel GPU.
Example on Intel GPU:
.. code-block:: python
from langchain_community.embeddings import IpexLLMBgeEmbeddings
embedding_model = IpexLLMBgeEmbeddings(
model_name="BAAI/bge-large-en-v1.5",
model_kwargs={"device": "xpu"},
encode_kwargs={"normalize_embeddings": True},
)
"""
client: Any #: :meta private:
model_name: str = DEFAULT_BGE_MODEL
"""Model name to use."""
cache_folder: Optional[str] = None
"""Path to store models.
Can be also set by SENTENCE_TRANSFORMERS_HOME environment variable."""
model_kwargs: Dict[str, Any] = Field(default_factory=dict)
"""Keyword arguments to pass to the model."""
encode_kwargs: Dict[str, Any] = Field(default_factory=dict)
"""Keyword arguments to pass when calling the `encode` method of the model."""
query_instruction: str = DEFAULT_QUERY_BGE_INSTRUCTION_EN
"""Instruction to use for embedding query."""
embed_instruction: str = ""
"""Instruction to use for embedding document."""
def __init__(self, **kwargs: Any):
"""Initialize the sentence_transformer."""
super().__init__(**kwargs)
try:
import sentence_transformers
from ipex_llm.transformers.convert import _optimize_post, _optimize_pre
except ImportError as exc:
base_url = (
"https://python.langchain.com/v0.1/docs/integrations/text_embedding/"
)
raise ImportError(
"Could not import ipex_llm or sentence_transformers. "
f"Please refer to {base_url}/ipex_llm/ "
"for install required packages on Intel CPU. "
f"And refer to {base_url}/ipex_llm_gpu/ "
"for install required packages on Intel GPU. "
) from exc
# Set "cpu" as default device
if "device" not in self.model_kwargs:
self.model_kwargs["device"] = "cpu"
if self.model_kwargs["device"] not in ["cpu", "xpu"]:
raise ValueError(
"IpexLLMBgeEmbeddings currently only supports device to be "
f"'cpu' or 'xpu', but you have: {self.model_kwargs['device']}."
)
self.client = sentence_transformers.SentenceTransformer(
self.model_name, cache_folder=self.cache_folder, **self.model_kwargs
)
# Add ipex-llm optimizations
self.client = _optimize_pre(self.client)
self.client = _optimize_post(self.client)
if self.model_kwargs["device"] == "xpu":
self.client = self.client.half().to("xpu")
if "-zh" in self.model_name:
self.query_instruction = DEFAULT_QUERY_BGE_INSTRUCTION_ZH
class Config:
"""Configuration for this pydantic object."""
extra = Extra.forbid
def embed_documents(self, texts: List[str]) -> List[List[float]]:
"""Compute doc embeddings using a HuggingFace transformer model.
Args:
texts: The list of texts to embed.
Returns:
List of embeddings, one for each text.
"""
texts = [self.embed_instruction + t.replace("\n", " ") for t in texts]
embeddings = self.client.encode(texts, **self.encode_kwargs)
return embeddings.tolist()
def embed_query(self, text: str) -> List[float]:
"""Compute query embeddings using a HuggingFace transformer model.
Args:
text: The text to embed.
Returns:
Embeddings for the text.
"""
text = text.replace("\n", " ")
embedding = self.client.encode(
self.query_instruction + text, **self.encode_kwargs
)
return embedding.tolist()