Dataset Card for NextGLab FAISS-v2.0

The NextGLab FAISS-v2.0 dataset is a dense vector index designed for Retrieval-Augmented Generation (RAG) tasks. It is constructed by embedding text from:

• srsRAN Project: A complete 5G RAN solution featuring an ORAN-native CU/DU.
• O-RAN Alliance specifications

Summary

• Total Documents: 88,808
• Total Words: 7,236,372
• Embedding Model: BAAI/bge-large-en-v1.5
• Vector Store: FAISS (Facebook AI Similarity Search)
• Developed by: NextG lab@ NC State
• License: MIT

The NextGLab FAISS-v2.0 dataset is fully compatible with all LLMs from the ORANSight-2.0 release, enabling seamless integration for Retrieval-Augmented Generation (RAG) workflows. We strongly recommend leveraging this dataset with ORANSight models to enhance contextual understanding, and code generation, optimize query-driven retrieval, and improve inference accuracy in O-RAN and telecom-related tasks.

How to Use the Dataset

Loading the FAISS Index

import os
from huggingface_hub import hf_hub_download
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS

# Create directory for FAISS files
os.makedirs("NextGLab_FAISS", exist_ok=True)

# Repository details
repo_id = "NextGLab/FAISS-v2.0"
index_filename = "index.faiss"
metadata_filename = "index.pkl"

# Download FAISS index and metadata
index_path = hf_hub_download(repo_id=repo_id, repo_type='dataset', filename=index_filename, local_dir="NextGLab_FAISS")
metadata_path = hf_hub_download(repo_id=repo_id, repo_type='dataset', filename=metadata_filename, local_dir="NextGLab_FAISS")

# Load FAISS index with embeddings
model_kwargs = {'device': 'cuda'}
embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-large-en-v1.5", model_kwargs=model_kwargs)
db = FAISS.load_local('NextGLab_FAISS', embeddings, allow_dangerous_deserialization=True)

Using FAISS as a Retriever

retriever = db.as_retriever(search_type="similarity", search_kwargs={"k": 5})

# Perform a retrieval
query = "What are the key components of an O-RAN Near-RT RIC?"
retrieved_docs = retriever.get_relevant_documents(query)
for doc in retrieved_docs:
    print(doc.page_content)

Citation

A detailed paper documenting the experiments and results achieved with this FAISS vector store will be available soon. Meanwhile, if you try this model, please cite the paper mentioned below to acknowledge the foundational work that enabled it.

@article{gajjar2024oran,
  title={Oran-bench-13k: An open source benchmark for assessing llms in open radio access networks},
  author={Gajjar, Pranshav and Shah, Vijay K},
  journal={arXiv preprint arXiv:2407.06245},
  year={2024}
}