U
h @ sV d dl mZmZmZmZ d dlmZ d dlmZm Z dZ
dZ dZ G dd deeZ
d S )
)AnyDictListOptional)
Embeddings) BaseModelFieldzBAAI/bge-small-en-v1.5z9Represent this question for searching relevant passages: u9 为这个句子生成表示以用于检索相关文章：c s e Zd ZU dZeed< eZeed< dZ e
e ed< e e dZ
eeef ed< e e dZeeef ed< eZeed < d
Zeed < ed fd
d ZG dd dZee eee dddZeee dddZ ZS )IpexLLMBgeEmbeddingsa 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 `_
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 `_
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
model_nameN cache_folder)default_factory model_kwargs
encode_kwargsquery_instruction embed_instruction)kwargsc
s t jf | z ddl}ddlm}m} W n> tk
rh } z d}td| d| d|W 5 d}~X Y nX d| jkr~d | jd< | jd d
krtd | jd d |j | j
fd
| j i| j| _ || j | _ || j | _ | jd dkr| j
d| _ d| j
kr
t| _dS )z$Initialize the sentence_transformer.r N)_optimize_post
_optimize_prezChttps://python.langchain.com/v0.1/docs/integrations/text_embedding/zDCould not import ipex_llm or sentence_transformers. Please refer to zD/ipex_llm/ for install required packages on Intel CPU. And refer to z;/ipex_llm_gpu/ for install required packages on Intel GPU. Zdevicecpu)r xpuzXIpexLLMBgeEmbeddings currently only supports device to be 'cpu' or 'xpu', but you have: .r r z-zh)super__init__sentence_transformersZ ipex_llm.transformers.convertr r ImportErrorr
ValueErrorZSentenceTransformerr r r
Zhalfto DEFAULT_QUERY_BGE_INSTRUCTION_ZHr )selfr r r r excbase_url __class__ K/tmp/pip-unpacked-wheel-9gdii04g/langchain_community/embeddings/ipex_llm.pyr C s>
z IpexLLMBgeEmbeddings.__init__c @ s e Zd ZdZdS )zIpexLLMBgeEmbeddings.ConfigZforbidN)__name__
__module__ __qualname__extrar% r% r% r& Configm s r+ )textsreturnc s, fdd|D } j j|f j}| S )zCompute doc embeddings using a HuggingFace transformer model.
Args:
texts: The list of texts to embed.
Returns:
List of embeddings, one for each text.
c s g | ]} j |d d qS )
)r replace).0tr r% r&
y s z8IpexLLMBgeEmbeddings.embed_documents..)r
encoder tolist)r r, Z
embeddingsr% r3 r& embed_documentsp s z$IpexLLMBgeEmbeddings.embed_documents)textr- c C s, | dd}| jj| j| f| j}| S )zCompute query embeddings using a HuggingFace transformer model.
Args:
text: The text to embed.
Returns:
Embeddings for the text.
r. r/ )r0 r
r5 r r r6 )r r8 Z embeddingr% r% r& embed_query} s z IpexLLMBgeEmbeddings.embed_query)r' r( r) __doc__r __annotations__DEFAULT_BGE_MODELr strr r r dictr r r DEFAULT_QUERY_BGE_INSTRUCTION_ENr r r r+ r floatr7 r9
__classcell__r% r% r# r& r s
# *
r N)typingr r r r Zlangchain_core.embeddingsr Zlangchain_core.pydantic_v1r r r< r? r r r% r% r% r& s