No article text was recorded for this version.
quantization-induced safety degradation
Version 1 of 128 papers50 conceptsfrozen 2026-10-04
Papers in this version
- Blog: Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compression
- Blog: PALMBENCH: A COMPREHENSIVE BENCHMARK OF COMPRESSED LARGE LANGUAGE MODELS ON MOBILE PLATFORMS
- Blog: Evaluating the Safety and Skill Reasoning of Large Reasoning Models Under Compute Constraints
- Blog: Toward Safe Quantization-Aware Fine-tuning: Understanding and Mitigating Safety Alignment Degradation
- Blog: Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Models
- Blog: Durable Quantization Conditioned Misalignment Attack on Large Language Models
- Blog: Mind the Gap: A Practical Attack on GGUF Quantization
- Blog: Q-realign: Piggybacking Realignment on Quantization for Safe and Efficient LLM Deployment
- Blog: HarmLevelBench: Evaluating Harm-Level Compliance and the Impact of Quantization on Model Alignment
- Blog: KVzip: Query-Agnostic KV Cache Compression with Context Reconstruction
- Blog: Beyond Perplexity: Multi-dimensional Safety Evaluation of LLM Compression
- Blog: A Comprehensive Evaluation of Quantization Strategies for Large Language Models
- Blog: 70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float (DFloat11)
- Blog: LLM-QAT: Data-Free Quantization Aware Training for Large Language Models
- Blog: Surprising Effectiveness of pretraining Ternary Language Model at Scale
- Blog: Mechanistically Interpreting Compression in Vision-Language Models
- Blog: BitDelta: Your Fine-Tune May Only Be Worth One Bit
- Blog: Reliability Scaling Laws for Quantized Large Language Models
- Blog: Certified Quantization Strategy Synthesis for Neural Networks
- Blog: Exploiting LLM Quantization
- Blog: The Channel Geometry of Refusal: Mechanistic Diagnosis of Alignment Collapse Under KV Quantization
- Blog: KV Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization
- Blog: RaBiT: Residual Aware Binarization Training for Accurate and Efficient LLMs
- Blog: NSNQuant: A Double Normalization Approach for Calibration-Free Low-Bit Vector Quantization of KV Cache
- Blog: Investigating the Impact of Quantization Methods on the Safety and Reliability of Large Language Models
- Blog: Benchmarking the Reliability of Post-training Quantization: a Particular Focus on Worst-case Performance
- Blog: Compressed but Compromised? A Study of Jailbreaking in Compressed LLMs
- Blog: ZipMoE: Efficient On-Device MoE Serving via Lossless Compression and Cache-Affinity Scheduling
History & citation
Current · Permanent link
- This direction split from The Fragility of Alignment Under Model Compression (79% overlap)
Cite this direction
Latest version (a living page):
@misc{lacuna_quantization_induced_safety_degradation_c26f0c17,
title = {quantization-induced safety degradation},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/quantization-induced-safety-degradation/rdr_c26f0c17c72e41d29b76559e8249b107},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}This exact version (version 1, frozen text):
@misc{lacuna_quantization_induced_safety_degradation_c26f0c17_v1,
title = {quantization-induced safety degradation},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/quantization-induced-safety-degradation/rdr_c26f0c17c72e41d29b76559e8249b107/v/1},
note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}