DSPy: The framework for programming—not prompting—language models
Python
fix(dspy): fixed bug in retriever part
Emanuele Fenocchi010d7ba93505
diff --git a/dspy/retrieve/deeplake_rm.py b/dspy/retrieve/deeplake_rm.pyindex 84bc7075d..df38fbefb 100644--- a/dspy/retrieve/deeplake_rm.py+++ b/dspy/retrieve/deeplake_rm.py@@ -11,15 +11,14 @@ import dspyfrom dsp.utils import dotdicttry:- import openai.error+ import openaiERRORS = (- openai.error.RateLimitError,- openai.error.ServiceUnavailableError,- openai.error.APIError,+ openai.RateLimitError,+ openai.APIError,)except Exception:- ERRORS = (openai.error.RateLimitError, openai.error.APIError)+ ERRORS = (openai.RateLimitError, openai.APIError)class DeeplakeRM(dspy.Retrieve):@@ -58,13 +57,15 @@ class DeeplakeRM(dspy.Retrieve):k: int = 3,):try:- from deeplake import VectorStore+ from deeplake import VectorStoreexcept ImportError:- raise ImportError(- "The 'deeplake' extra is required to use DeepLakeRM. Install it with `pip install dspy-ai[deeplake]`",- )+ raise ImportError("The 'deeplake' extra is required to use DeepLakeRM. Install it with `pip install dspy-ai[deeplake]`",)+self._deeplake_vectorstore_name = deeplake_vectorstore_name- self._deeplake_client = deeplake_client+ self._deeplake_client = deeplake_client(+ path=self._deeplake_vectorstore_name,+ embedding_function=self.embedding_function,+ )super().__init__(k=k)@@ -73,13 +74,11 @@ class DeeplakeRM(dspy.Retrieve):texts = [texts]texts = [t.replace("\n", " ") for t in texts]- return [- data["embedding"]- for data in openai.Embedding.create(input=texts, model=model)["data"]- ]-++ return [data.embedding for data in openai.embeddings.create(input = texts, model=model).data]+def forward(- self, query_or_queries: Union[str, List[str]], k: Optional[int],**kwargs,+ self, query_or_queries: Union[str, List[str]], k: Optional[int],) -> dspy.Prediction:"""Search with DeepLake for self.k top passages for query@@ -103,10 +102,7 @@ class DeeplakeRM(dspy.Retrieve):passages = defaultdict(float)#deeplake doesn't support batch querying, manually querying each query and storing themfor query in queries:- results = self._deeplake_client(- path=self._deeplake_vectorstore_name,- embedding_function=self.embedding_function,- ).search(query, k=k,**kwargs)+ results = self._deeplake_client.search(query, k=k)for score,text in zip(results.get('score',0.0),results.get('text',"")):passages[text] += score