Utilities

RESK_Embedder and SimpleEmbedder

from resk_llm.utilities.resk_embedding_utils import RESK_Embedder, SimpleEmbedder

simple = SimpleEmbedder(dimension=64)
vec = simple.embed("hello world")

# sklearn-based (requires scikit-learn)
# emb = RESK_Embedder(dimension=128)
# emb.train(["a corpus", "of texts"]) 
# v = emb.embed("hello world")

RESK_TextAnalyzer

from resk_llm.utilities.resk_text_analysis import RESK_TextAnalyzer

ta = RESK_TextAnalyzer()
print(ta.clean_text("foo\u200Bbar"))

RESK_VectorDatabase

See Detectors page for end-to-end example, including external DBs.

---

title: Utilities

Text analysis

resk_llm.utilities.resk_text_analysis.RESK_TextAnalyzer detects homoglyphs, invisible chars, etc.

Embeddings and vector DB

RESK_Embedder and RESK_VectorDatabase help store and compare embeddings for similarity to known attacks.

from resk_llm.utilities.resk_vector_db import RESK_VectorDatabase
db = RESK_VectorDatabase(embedding_dim=1536, similarity_threshold=0.85)