github/langchain
1
0
Fork 0
mirror of https://github.com/hwchase17/langchain.git synced 2025-07-16 17:53:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

Improve query prompt of BGE embeddings (#8908)

Browse Source 
Replace this comment with:
- Description: Improved query of BGE embeddings after talking with the
devs of BGE embeddings ,
  - Dependencies: any dependencies required for this change,
  - Tag maintainer: @hwchase17 ,
  - Twitter handle: @ManabChetia3

---------

Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
This commit is contained in:
manmax31 2023-08-09 03:20:37 +10:00 committed by GitHub
parent c5c0735fc4
commit 4a07fba9f0
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 12 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
libs/langchain/langchain/embeddings/huggingface.py
View File

@ -11,12 +11,10 @@ DEFAULT_EMBED_INSTRUCTION = "Represent the document for retrieval: "
DEFAULT_QUERY_INSTRUCTION = ( DEFAULT_QUERY_INSTRUCTION = (
"Represent the question for retrieving supporting documents: " "Represent the question for retrieving supporting documents: "
) )
DEFAULT_EMBED_BGE_INSTRUCTION = ( DEFAULT_QUERY_BGE_INSTRUCTION_EN = (
"Represent this sentence for searching relevant passages: "
)
DEFAULT_QUERY_BGE_INSTRUCTION = (
"Represent this question for searching relevant passages: " "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 HuggingFaceEmbeddings(BaseModel, Embeddings): class HuggingFaceEmbeddings(BaseModel, Embeddings):
@ -190,7 +188,7 @@ class HuggingFaceBgeEmbeddings(BaseModel, Embeddings):
model_name = "BAAI/bge-large-en" model_name = "BAAI/bge-large-en"
model_kwargs = {'device': 'cpu'} model_kwargs = {'device': 'cpu'}
encode_kwargs = {'normalize_embeddings': False} encode_kwargs = {'normalize_embeddings': True}
hf = HuggingFaceBgeEmbeddings( hf = HuggingFaceBgeEmbeddings(
model_name=model_name, model_name=model_name,
model_kwargs=model_kwargs, model_kwargs=model_kwargs,
@ -208,9 +206,7 @@ class HuggingFaceBgeEmbeddings(BaseModel, Embeddings):
"""Key word arguments to pass to the model.""" """Key word arguments to pass to the model."""
encode_kwargs: Dict[str, Any] = Field(default_factory=dict) encode_kwargs: Dict[str, Any] = Field(default_factory=dict)
"""Key word arguments to pass when calling the `encode` method of the model.""" """Key word arguments to pass when calling the `encode` method of the model."""
embed_instruction: str = DEFAULT_EMBED_BGE_INSTRUCTION query_instruction: str = DEFAULT_QUERY_BGE_INSTRUCTION_EN
"""Instruction to use for embedding documents."""
query_instruction: str = DEFAULT_QUERY_BGE_INSTRUCTION
"""Instruction to use for embedding query.""" """Instruction to use for embedding query."""
def __init__(self, **kwargs: Any): def __init__(self, **kwargs: Any):
@ -228,6 +224,8 @@ class HuggingFaceBgeEmbeddings(BaseModel, Embeddings):
self.client = sentence_transformers.SentenceTransformer( self.client = sentence_transformers.SentenceTransformer(
self.model_name, cache_folder=self.cache_folder, **self.model_kwargs self.model_name, cache_folder=self.cache_folder, **self.model_kwargs
) )
if "-zh" in self.model_name:
self.query_instruction = DEFAULT_QUERY_BGE_INSTRUCTION_ZH
class Config: class Config:
"""Configuration for this pydantic object.""" """Configuration for this pydantic object."""
@ -243,8 +241,8 @@ class HuggingFaceBgeEmbeddings(BaseModel, Embeddings):
Returns: Returns:
List of embeddings, one for each text. List of embeddings, one for each text.
""" """
instruction_pairs = [[self.embed_instruction, text] for text in texts] texts = [t.replace("\n", " ") for t in texts]
embeddings = self.client.encode(instruction_pairs, **self.encode_kwargs) embeddings = self.client.encode(texts, **self.encode_kwargs)
return embeddings.tolist() return embeddings.tolist()
def embed_query(self, text: str) -> List[float]: def embed_query(self, text: str) -> List[float]:
@ -256,6 +254,8 @@ class HuggingFaceBgeEmbeddings(BaseModel, Embeddings):
Returns: Returns:
Embeddings for the text. Embeddings for the text.
""" """
instruction_pair = [self.query_instruction, text] text = text.replace("\n", " ")
embedding = self.client.encode([instruction_pair], **self.encode_kwargs)[0] embedding = self.client.encode(
self.query_instruction + text, **self.encode_kwargs
)
return embedding.tolist() return embedding.tolist()