Update QdrantRag.py

QdrantRag.py

@@ -21,10 +21,6 @@ qdrant_client = QdrantClient(url=os.getenv("qdrant_url"), api_key=os.getenv("qdr
 sparse_encoder = SparseTextEmbedding(model_name="prithivida/Splade_PP_en_v1")
 co = cohere.ClientV2(os.getenv("cohere_api_key"))
 
-dataset = load_dataset("Karbo31881/Pokemon_images")
-ds = dataset["train"]
-labels = ds["text"]
-
 def get_sparse_embedding(text: str, model: SparseTextEmbedding):
     embeddings = list(model.embed(text))
     vector = {f"sparse-text": models.SparseVector(indices=embeddings[0].indices, values=embeddings[0].values)}
@@ -169,44 +165,3 @@ class NeuralSearcher:
         )
         payloads = [hit.payload["label"] for hit in search_result]
         return payloads
-
-qdrant_client.recreate_collection(
-    collection_name="pokemon_texts",
-    vectors_config={"dense-text": models.VectorParams(
-        size=768,  # Vector size is defined by used model
-        distance=models.Distance.COSINE,
-    )},
-    sparse_vectors_config={"sparse-text": models.SparseVectorParams(
-        index=models.SparseIndexParams(
-            on_disk=False
-        )
-    )}
-)
-textdata = load_dataset("wanghaofan/pokemon-wiki-captions")
-names = textdata["train"]["name_en"]
-texts = textdata["train"]["text_en"]
-
-c = 0
-
-for j in range(len(texts)):
-    txt = names[j].upper() + "\n\n" + texts[j]
-    l = c+1
-    upload_text_to_qdrant(qdrant_client, "pokemon_texts", encoder, txt, c, l)
-    c = l+1
-
-qdrant_client.recreate_collection(
-    collection_name="pokemon_images",
-    vectors_config=models.VectorParams(
-        size=1024,  # Vector size is defined by used model
-        distance=models.Distance.COSINE,
-    ),
-)
-upload_images_to_qdrant(qdrant_client, "pokemon_images", "data/vector_pokemon.npy", labels)
-
-qdrant_client.recreate_collection(
-    collection_name="semantic_cache",
-    vectors_config=models.VectorParams(
-        size=768,  # Vector size is defined by used model
-        distance=models.Distance.COSINE,
-    ),
-)