홈 › Papers

Granite Guardian

2024-12-10 · Inkit Padhi, Manish Nagireddy, Giandomenico Cornacchia, Subhajit Chaudhury, Tejaswini Pedapati, Pierre Dognin, Keerthiram Murugesan, Erik Miehling, Martín Santillán Cooper, Kieran Fraser, Giulio Zizzo, Muhammad Zaid Hameed, Mark Purcell, Michael Desmond, Qian Pan, Zahra Ashktorab, Inge Vejsbjerg, Elizabeth M. Daly, Michael Hind, Werner Geyer, Ambrish Rawat, Kush R. Varshney, Prasanna Sattigeri

We introduce the Granite Guardian models, a suite of safeguards designed to provide risk detection for prompts and responses, enabling safe and responsible use in combination with any large language model (LLM). These models offer comprehensive coverage across multiple risk dimensions, including social bias, profanity, violence, sexual content, unethical behavior, jailbreaking, and hallucination-related risks such as context relevance, groundedness, and answer relevance for retrieval-augmented generation (RAG). Trained on a unique dataset combining human annotations from diverse sources and synthetic data, Granite Guardian models address risks typically overlooked by traditional risk detection models, such as jailbreaks and RAG-specific issues. With AUC scores of 0.871 and 0.854 on harmful content and RAG-hallucination-related benchmarks respectively, Granite Guardian is the most generalizable and competitive model available in the space. Released as open-source, Granite Guardian aims to promote responsible AI development across the community. https://github.com/ibm-granite/granite-guardian

📄 PDF Abstract BibTeX arXiv:2412.07724

Code (1)

ibm-granite/granite-guardian 공식 구현

Tasks

HallucinationLanguage ModelingLanguage ModellingLarge Language ModelRAGRetrieval-augmented Generation

When Safety Geometry Collapses: Fine-Tuning Vulnerabilities in Agentic Guard Models

2026-04-08 · Ismail Hossain, Sai Puppala, Jannatul Ferdaus, Md Jahangir Alam 외 arxiv

A guard model fine-tuned on entirely benign data can lose all safety alignment -- not through adversarial manipulation, but through standard domain specialization. We demonstrate this failure across three purpose-built s…

AprielGuard

2025-12-23 · Jaykumar Kasundra, Anjaneya Praharaj, Sourabh Surana, Lakshmi Sirisha Chodisetty 외 arxiv

Safeguarding large language models (LLMs) against unsafe or adversarial behavior is critical as they are increasingly deployed in conversational and agentic settings. Existing moderation tools often treat safety risks (e…

Identifying Features Associated with Bias Against 93 Stigmatized Groups in Language Models and Guardrail Model Safety Mitigation

2025-12-22 · Anna-Maria Gueorguieva, Aylin Caliskan arxiv

Large language models (LLMs) have been shown to exhibit social bias, however, bias towards non-protected stigmatized identities remain understudied. Furthermore, what social features of stigmas are associated with bias i…

HalluGuard: Evidence-Grounded Small Reasoning Models to Mitigate Hallucinations in Retrieval-Augmented Generation

2025-10-01 · Loris Bergeron, Ioana Buhnila, Jérôme François, Radu State arxiv

Large Language Models (LLMs) excel in many NLP tasks but remain prone to hallucinations, limiting trust in real-world applications. We present HalluGuard, a 4B-parameter Small Reasoning Model (SRM) for mitigating halluci…

Granite-speech: open-source speech-aware LLMs with strong English ASR capabilities

2025-05-13 · George Saon, Avihu Dekel, Alexander Brooks, Tohru Nagano 외

Granite-speech LLMs are compact and efficient speech language models specifically designed for English ASR and automatic speech translation (AST). The models were trained by modality aligning the 2B and 8B parameter vari…

automatic-speech-translation Benchmarking