huawei-csl/AC-LoRA
Welcome to the official repository of AC-LORA: (Almost) Training-Free Access Control-Aware Multi-Modal LLMs, a mechanism that provides training-free secure access control for LLMs using separate LoRAs fine-tuned with sensitive data, and merge them based on the user query and permission.
This project helps organizations use large language models (LLMs) with sensitive, proprietary information while maintaining strict access control. It takes an LLM and separate, secure data sets (like internal documents or project details) and produces an LLM that can answer questions using the permitted data for each user, without risking data leaks. This is ideal for businesses, government agencies, or research institutions who want to leverage LLMs internally without compromising data security.
Use this if you need to deploy an internal LLM that draws upon different confidential datasets, and you must ensure only authorized users can access specific pieces of information.
Not ideal if your LLM only uses public data or if fine-grained access control on sensitive internal information is not a primary concern.
Stars
20
Forks
1
Language
Jupyter Notebook
License
CC0-1.0
Category
Last pushed
Nov 14, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/huawei-csl/AC-LoRA"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
axolotl-ai-cloud/axolotl
Go ahead and axolotl questions
google/paxml
Pax is a Jax-based machine learning framework for training large scale models. Pax allows for...
JosefAlbers/PVM
Phi-3.5 for Mac: Locally-run Vision and Language Models for Apple Silicon
iamarunbrahma/finetuned-qlora-falcon7b-medical
Finetuning of Falcon-7B LLM using QLoRA on Mental Health Conversational Dataset
h2oai/h2o-wizardlm
Open-Source Implementation of WizardLM to turn documents into Q:A pairs for LLM fine-tuning