add bge-small-finetuned and bge-large-finetunes chromadb collections

app/src/chroma.py

@@ -4,9 +4,9 @@ from chromadb import Collection, QueryResult
 from chromadb.api import ClientAPI
 from pandas import DataFrame
 from src.sentence import sentence_embed
-from src.settings import CHROMA_COLLECTION, CHROMA_DIR
+from src.settings import CHROMA_COLLECTION, CHROMA_DB
 
-chroma_client: ClientAPI = chromadb.PersistentClient(path=CHROMA_DIR)
+chroma_client: ClientAPI = chromadb.PersistentClient(path=CHROMA_DB)
 chroma_collection: Collection = chroma_client.get_or_create_collection(
     name=CHROMA_COLLECTION, metadata={"hnsw:space": "cosine"}
 )

app/src/settings.py

@@ -8,9 +8,10 @@ load_dotenv(find_dotenv())
 
 SRC_DIR: str = os.path.dirname(__file__)
 DATA_DIR: str = os.path.join(SRC_DIR, "../../data")
-CHROMA_DIR: str = os.path.join(SRC_DIR, "../../.chroma")
 MODELS_DIR: str = os.path.join(SRC_DIR, "../../models")
-CHROMA_COLLECTION: str = "bge_small_finetuned_astra_collection_10k"
+CHROMA_DIR: str = os.path.join(SRC_DIR, "../../.chroma")
+CHROMA_DB: str = os.path.join(CHROMA_DIR, "bge-small-finetuned-chroma")
+CHROMA_COLLECTION: str =  "bge_small_finetuned_astra_collection" # use bge_small_finetuned_astra_collection for the bege-large model and embeddings
 
 
 class Settings(BaseSettings):