working local version

RagPipeline.py

@@ -23,7 +23,7 @@ class RetrievalAugmentedQAPipeline:
         context_list = self.vector_db_retriever.search_by_text(user_query, k=4)
 
         context_prompt = ""
-        for context in context_list[0]:
+        for context in context_list:
             context_prompt += context[0] + "\n"
 
         formatted_system_prompt = self.system_role_prompt.create_message()

aimakerspace/vectordatabase.py

@@ -6,7 +6,7 @@ from typing import List, Tuple, Callable
 from aimakerspace.openai_utils.embedding import EmbeddingModel
 import asyncio
 from qdrant_client import models, QdrantClient
-from qdrant_client.models import PointStruct,VectorParams,Distance
+from qdrant_client.models import PointStruct,VectorParams,Distance,Batch,VectorStruct,Payload
 
 collection_name = "embedding_collection"
 
@@ -76,20 +76,20 @@ class VectorDatabase:
             self.qdrant_client = QdrantClient(":memory:")
             vector_params = VectorParams(
                 size=embedding_model.dimensions,  # vector size 
-                distance="Cosine"  # distance metric
+                distance=Distance.COSINE
             )
-            self.qdrant_client.recreate_collection(
+            self.qdrant_client.create_collection(
                 collection_name=collection_name,
-                vectors_config={"default": vector_params},
+                vectors_config={"text": vector_params},
             )
 
-    def insert(self, key: str, vector: np.array) -> None:
+    def insert(self, key: str, vectors: np.array) -> None:
         idx = str(uuid.uuid4())
         payload = {"text": key}  
 
         point = PointStruct(
             id=idx,
-            vector={"default": vector.tolist()},
+            vector={"default": vectors.tolist()},
             payload=payload
         )
         # Insert the vector into Qdrant with the associated document
@@ -97,9 +97,25 @@ class VectorDatabase:
             collection_name=collection_name,
             points=[point]
         )
-        print(f"Inserted vector with ID {idx}: {vector}")
-        
-
+        # print(f"Inserted vector with ID {idx}: {vector}")
+        # self.qdrant_client.upsert(
+        #     collection_name=collection_name,
+        #     points= [
+        #         [PointStruct(
+        #             id=idx,
+        #             vector=vector,
+        #             payload={"text": key}
+        #         )]
+        #         for idx, vector in enumerate(vectors)
+        #     ])
+        # self.qdrant_client.add(
+        #     collection_name=collection_name,
+        #     documents=[key],
+        #     metadata=[],
+        #     ids=str(uuid.uuid4())
+        # )
+
+            
     def search(
         self,
         query_vector: np.array,
@@ -114,7 +130,7 @@ class VectorDatabase:
 
         search_results = self.qdrant_client.search(
             collection_name=collection_name,
-            query_vector=query_vector,
+            query_vector=('text',query_vector),
             limit=k
         )
         return [(result.payload['text'], result.score) for result in search_results]
@@ -136,8 +152,25 @@ class VectorDatabase:
 
     async def abuild_from_list(self, list_of_text: List[str]) -> "VectorDatabase":
         embeddings = await self.embedding_model.async_get_embeddings(list_of_text)
-        for text, embedding in zip(list_of_text, embeddings):
-            self.insert(text, np.array(embedding))
+        # vs = VectorStruct()
+        # VectorStruct = Union[
+        #     List[StrictFloat],
+        #     List[List[StrictFloat]],
+        #     Dict[StrictStr, Vector],
+        points = [
+            models.PointStruct(
+                id=str(uuid.uuid4()),
+                vector={ 'text': embedding},
+                payload={
+                    "text": text
+                    }
+            )
+            for text, embedding in zip(list_of_text, embeddings)
+        ]
+        self.qdrant_client.upsert(
+            collection_name=collection_name,
+            points=points
+        )
         return self