common.py

# Embedding model name from HuggingFace
EMBEDDING_MODEL_NAME = "sentence-transformers/all-mpnet-base-v2"

# Embedding model kwargs
MODEL_KWARGS = {"device": "cpu"}  # or "cuda"

# The similarity threshold in %
# where 1.0 is 100% "known threat" from the database.
# Any vectors found above this value will teigger an anomaly on the provided prompt.
SIMILARITY_ANOMALY_THRESHOLD = 0.1

# Number of prompts to retreive (TOP K)
K = 3

# Number of similar prompts to revreive before choosing TOP K
FETCH_K = 5

# The path of the dataset (to ingest)
DATASET_DIR = "/code/data/datasets/"
# DATASET_DIR = "../data/datasets/"
# DATASET_DIR = "/code/data/datasets/malicious_prompts/"
# DATASET_DIR = "../data/datasets/malicious_prompts/"
# DATASET_DIR = "/code/data/datasets/new_prompts_08_2024"


# The path of the vector store
# VECTORSTORE_FILENAME = "vectorstores/all"
# VECTORSTORE_FILENAME = "vectorstores/jailbreak_v_28k"
# VECTORSTORE_FILENAME = "vectorstores/malicious_prompts"
# VECTORSTORE_FILENAME = "vectorstores/malicious_prompts_chumk_200_overlap_100"
# VECTORSTORE_FILENAME = "vectorstores/malicious_prompts_chumk_200_overlap_100_2"
# VECTORSTORE_FILENAME = (
#     "/code/data/vectorstores/malicious_prompts_knowledge_base"
# )
# VECTORSTORE_FILENAME = "../data/vectorstores/all_prompts"
# VECTORSTORE_FILENAME = "../data/vectorstores/best"
# VECTORSTORE_FILENAME = "/code/data/vectorstores/pliny-the-prompter-prompts"
VECTORSTORE_FILENAME = "/code/vectorstore"
# VECTORSTORE_FILENAME = "../data/vectorstores/09-2024"

# VECTORSTORE_FILENAME = "/code/data/vectorstores/best"