Skip to content

armgit / dspy

Public Imported from GitHub
0

DSPy: The framework for programming—not prompting—language models

Python

kwargs parameter for retrievers

Athe-kunaldc6c53985d02
diff --git a/dsp/primitives/search.py b/dsp/primitives/search.py
index 1ca4b966d..aef9cd8eb 100644
--- a/dsp/primitives/search.py
+++ b/dsp/primitives/search.py
@@ -24,13 +24,13 @@ def retrieve(query: str, k: int, **kwargs) -> list[str]:
return passages
-def retrieveRerankEnsemble(queries: list[str], k: int) -> list[str]:
+def retrieveRerankEnsemble(queries: list[str], k: int,**kwargs) -> list[str]:
if not (dsp.settings.rm and dsp.settings.reranker):
raise AssertionError("Both RM and Reranker are needed to retrieve & re-rank.")
queries = [q for q in queries if q]
passages = {}
for query in queries:
- retrieved_passages = dsp.settings.rm(query, k=k*3)
+ retrieved_passages = dsp.settings.rm(query, k=k*3,**kwargs)
passages_cs_scores = dsp.settings.reranker(query, [psg.long_text for psg in retrieved_passages])
for idx in np.argsort(passages_cs_scores)[::-1]:
psg = retrieved_passages[idx]
@@ -42,7 +42,7 @@ def retrieveRerankEnsemble(queries: list[str], k: int) -> list[str]:
return [text for _, text in sorted(passages, reverse=True)[:k]]
-def retrieveEnsemble(queries: list[str], k: int, by_prob: bool = True) -> list[str]:
+def retrieveEnsemble(queries: list[str], k: int, by_prob: bool = True,**kwargs) -> list[str]:
"""Retrieves passages from the RM for each query in queries and returns the top k passages
based on the probability or score.
"""
@@ -58,7 +58,7 @@ def retrieveEnsemble(queries: list[str], k: int, by_prob: bool = True) -> list[s
passages = {}
for q in queries:
- for psg in dsp.settings.rm(q, k=k * 3):
+ for psg in dsp.settings.rm(q, k=k * 3,**kwargs):
if by_prob:
passages[psg.long_text] = passages.get(psg.long_text, 0.0) + psg.prob
else:
diff --git a/dspy/retrieve/chromadb_rm.py b/dspy/retrieve/chromadb_rm.py
index 1b5de8874..cabb0e5d6 100644
--- a/dspy/retrieve/chromadb_rm.py
+++ b/dspy/retrieve/chromadb_rm.py
@@ -122,7 +122,7 @@ class ChromadbRM(dspy.Retrieve):
return self.ef(queries)
def forward(
- self, query_or_queries: Union[str, List[str]], k: Optional[int] = None,
+ self, query_or_queries: Union[str, List[str]], k: Optional[int] = None, **kwargs
) -> dspy.Prediction:
"""Search with db for self.k top passages for query
@@ -142,7 +142,7 @@ class ChromadbRM(dspy.Retrieve):
k = self.k if k is None else k
results = self._chromadb_collection.query(
- query_embeddings=embeddings, n_results=k,
+ query_embeddings=embeddings, n_results=k,**kwargs
)
zipped_results = zip(
diff --git a/dspy/retrieve/clarifai_rm.py b/dspy/retrieve/clarifai_rm.py
index 654234d2c..8f4ac659d 100644
--- a/dspy/retrieve/clarifai_rm.py
+++ b/dspy/retrieve/clarifai_rm.py
@@ -57,7 +57,7 @@ class ClarifaiRM(dspy.Retrieve):
return requested_text
def forward(
- self, query_or_queries: Union[str, List[str]], k: Optional[int] = None,
+ self, query_or_queries: Union[str, List[str]], k: Optional[int] = None,**kwargs
) -> dspy.Prediction:
"""Uses clarifai-python SDK search function and retrieves top_k similar passages for given query,
Args:
@@ -85,7 +85,7 @@ class ClarifaiRM(dspy.Retrieve):
queries = [q for q in queries if q]
for query in queries:
- search_response = self.clarifai_search.query(ranks=[{"text_raw": query}])
+ search_response = self.clarifai_search.query(ranks=[{"text_raw": query}],**kwargs)
# Retrieve hits
hits = [hit for data in search_response for hit in data.hits]
diff --git a/dspy/retrieve/deeplake_rm.py b/dspy/retrieve/deeplake_rm.py
index 235108a91..1795d788d 100644
--- a/dspy/retrieve/deeplake_rm.py
+++ b/dspy/retrieve/deeplake_rm.py
@@ -79,7 +79,7 @@ class DeeplakeRM(dspy.Retrieve):
]
def forward(
- self, query_or_queries: Union[str, List[str]], k: Optional[int],
+ self, query_or_queries: Union[str, List[str]], k: Optional[int],**kwargs
) -> dspy.Prediction:
"""Search with DeepLake for self.k top passages for query
@@ -106,7 +106,7 @@ class DeeplakeRM(dspy.Retrieve):
results = self._deeplake_client(
path=self._deeplake_vectorstore_name,
embedding_function=self.embedding_function,
- ).search(query, k=k)
+ ).search(query, k=k,**kwargs)
for score,text in zip(results.get('score',0.0),results.get('text',"")):
passages[text] += score
diff --git a/dspy/retrieve/faiss_rm.py b/dspy/retrieve/faiss_rm.py
index f321d8736..8cd98466f 100755
--- a/dspy/retrieve/faiss_rm.py
+++ b/dspy/retrieve/faiss_rm.py
@@ -107,7 +107,7 @@ class FaissRM(dspy.Retrieve):
logging.debug(f" Hit {j} = {indices[j]}/{distances[j]}: {self._document_chunks[indices[j]]}")
return
- def forward(self, query_or_queries: Union[str, list[str]], k: Optional[int] = None) -> dspy.Prediction:
+ def forward(self, query_or_queries: Union[str, list[str]], k: Optional[int] = None,**kwargs) -> dspy.Prediction:
"""Search the faiss index for k or self.k top passages for query.
Args:
@@ -127,7 +127,7 @@ class FaissRM(dspy.Retrieve):
passages = [(self._document_chunks[ind], ind) for ind in index_list[0]]
return [dotdict({"long_text": passage[0], "index": passage[1]}) for passage in passages]
- distance_list, index_list = self._faiss_index.search(emb_npa, (k or self.k) * 3)
+ distance_list, index_list = self._faiss_index.search(emb_npa, (k or self.k) * 3,**kwargs)
# self._dump_raw_results(queries, index_list, distance_list)
passage_scores = {}
for emb in range(len(embeddings)):
diff --git a/dspy/retrieve/qdrant_rm.py b/dspy/retrieve/qdrant_rm.py
index 5c2af050b..1ee63f275 100644
--- a/dspy/retrieve/qdrant_rm.py
+++ b/dspy/retrieve/qdrant_rm.py
@@ -53,7 +53,7 @@ class QdrantRM(dspy.Retrieve):
super().__init__(k=k)
- def forward(self, query_or_queries: Union[str, List[str]], k: Optional[int]) -> dspy.Prediction:
+ def forward(self, query_or_queries: Union[str, List[str]], k: Optional[int],**kwargs) -> dspy.Prediction:
"""Search with Qdrant for self.k top passages for query
Args:
@@ -71,7 +71,7 @@ class QdrantRM(dspy.Retrieve):
k = k if k is not None else self.k
batch_results = self._qdrant_client.query_batch(
- self._qdrant_collection_name, query_texts=queries, limit=k)
+ self._qdrant_collection_name, query_texts=queries, limit=k,**kwargs)
passages_scores = defaultdict(float)
for batch in batch_results:
diff --git a/dspy/retrieve/retrieve.py b/dspy/retrieve/retrieve.py
index 71b4f5734..7f026e2aa 100644
--- a/dspy/retrieve/retrieve.py
+++ b/dspy/retrieve/retrieve.py
@@ -29,14 +29,14 @@ class Retrieve(Parameter):
def __call__(self, *args, **kwargs):
return self.forward(*args, **kwargs)
- def forward(self, query_or_queries: Union[str, List[str]], k: Optional[int] = None) -> Prediction:
+ def forward(self, query_or_queries: Union[str, List[str]], k: Optional[int] = None,**kwargs) -> Prediction:
queries = [query_or_queries] if isinstance(query_or_queries, str) else query_or_queries
queries = [query.strip().split('\n')[0].strip() for query in queries]
# print(queries)
# TODO: Consider removing any quote-like markers that surround the query too.
k = k if k is not None else self.k
- passages = dsp.retrieveEnsemble(queries, k=k)
+ passages = dsp.retrieveEnsemble(queries, k=k,**kwargs)
return Prediction(passages=passages)
# TODO: Consider doing Prediction.from_completions with the individual sets of passages (per query) too.