add revision

load_data.py

@@ -24,6 +24,7 @@ load_dotenv()
 
 HF_TOKEN = os.getenv("HF_TOKEN")
 EMBEDDING_MODEL_NAME = "Snowflake/snowflake-arctic-embed-m-long"
+EMBEDDING_MODEL_REVISION = "ac9d0cb43661ee1f7d67b3aa63614d65a6c86463"
 INFERENCE_MODEL_URL = (
     "https://pqzap00ebpl1ydt4.us-east-1.aws.endpoints.huggingface.cloud"
 )
@@ -49,7 +50,9 @@ def get_chroma_client():
 def get_embedding_function():
     logger.info(f"Initializing embedding function with model: {EMBEDDING_MODEL_NAME}")
     return embedding_functions.SentenceTransformerEmbeddingFunction(
-        model_name=EMBEDDING_MODEL_NAME, trust_remote_code=True
+        model_name=EMBEDDING_MODEL_NAME,
+        trust_remote_code=True,
+        revision=EMBEDDING_MODEL_REVISION,
     )