added EvalDataset Generation

app.py

@@ -238,7 +238,7 @@ class BSIChatbot:
             #newprint("Saving Embeddings took", end-start, "seconds!")
         else:
             start = time.time()
-            if vectorstore == None:
+            if vectorstore is None:
                 vectorstore = FAISS.load_local(self.embedPath, self.embedding_model, allow_dangerous_deserialization=True)
             #self.vectorstore.index = index_gpu
             end = time.time()
@@ -280,7 +280,7 @@ class BSIChatbot:
         #print(vectorstore.index_to_docstore_id)
         #newprint(vectorstore)
         # Iteriere über alle IDs im index_to_docstore_id
-        if docstore == None:
+        if docstore is None:
             docstore = vectorstore.docstore._dict.values()
 
         #for doc_id in vectorstore.index_to_docstore_id.values():
@@ -336,14 +336,15 @@ class BSIChatbot:
         global rerankingModel
         if hybridSearch == True:
             allDocs = self.retrieveDocFromFaiss()
-            if bm25_retriever == None:
+            if bm25_retriever is None:
                 bm25_retriever = BM25Retriever.from_documents(allDocs)
             #TODO!
             retriever_k=15
             bm25_retriever.k= retriever_k
             vectordb = vectorstore.as_retriever(search_kwargs={"k":retriever_k})
             ensemble_retriever = EnsembleRetriever(retrievers=[bm25_retriever, vectordb], weights=[0.5, 0.5])
-            retrieved_chunks = ensemble_retriever.get_relevant_documents(query)
+            retrieved_chunks = ensemble_retriever.invoke(query)
+            #retrieved_chunks = ensemble_retriever.get_relevant_documents(query)
             #newprint("DBG: Number of Chunks retrieved")
             #newprint(len(retrieved_chunks))
         else:
@@ -364,7 +365,7 @@ class BSIChatbot:
             i = i + 1
 
         if rerankingStep == True:
-            if rerankingModel == None:
+            if rerankingModel is None:
                 print("initializing Reranker-Model..")
                 self.initializeRerankingModel()
             print("Starting Reranking Chunks...")