GraphCanon updated Oct 5, 2026 · GitHub synced Oct 5, 2026
43views this month
Decision brief
IB4LLMs (IBProtector) is an LLM jailbreak defense method using the Information Bottleneck principle to prevent adversarial prompts while preserving key information.
Good fit when
- When you need a specialized tool for guarding against jailbreaks in your language models without losing important data.
- For researchers or practitioners focused on mitigating adversarial attacks specifically through the lens of information bottlenecks.
Avoid when
- If you require a more generalized model protection approach that does not rely strictly on the Information Bottleneck principle.
- When your environment lacks support for specific packages like `fschat==0.2.20` which is crucial and cannot be updated due to potential conflicts.
- Pricing:
- unknown - License information unavailable; specific model pricing not provided.
- Requirements:
- Dependencies include datasets==2.14.5, torch==2.1.1, transformers==4.40.1 among others.
Observed Jul 12, 2026 · Source: enrich:decision_facts
Verify the decision
Maintenance and security
Full trust report- Maintenance
- Dormant (696d since push)
- As of Oct 5, 2026
- Provenance
- Not a fork · Personal account
- As of Oct 5, 2026
- Security (OSV)
- 77 low (77 low)
- As of Jul 11, 2026
Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.
Install
pip install IB4LLMsSimilar tools
Same-category neighbours. No typed graph edges are catalogued for this tool yet.
Evidence and technical details
Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.
Overview
Proposes IBProtector, an LLM jailbreak defense method based on the Information Bottleneck principle.
Capability facts
- Languages
- python
Source: github.language · Oct 5, 2026
Categories
Compatibility
Sourced claims from the README excerpt - not unsourced marketing copy.
Source: README excerpt (regex_v1, Oct 5, 2026)
python test_finetuning.pySource link
Tags
README
<p align=center :fire: </p [Arxiv Paper] [Slides] [中文版] [Website Page] TL;DR : We propose IBProtector, the first LLM jailbreak defending method based on the Information Bottleneck principle. Our protector efficiently defends against adversarial prompts without losing key informat...
For agents
This page has a .md twin and JSON over the API.