In "Compress or Route?" (Johnson, 2026), we found that code generation tolerates aggressive prompt compression (r >= 0.6) while chain-of-thought reasoning degrades gradually. That study was limited to HumanEval (164 problems), left the "perplexity paradox" mechanism unvalidated, and provided no adaptive algorithm. This paper addresses all three gaps. First, we validate across six code benchmarks (HumanEval, MBPP, HumanEval+, MultiPL-E) and four reasoning benchmarks (GSM8K, MATH, ARC-Challenge, MMLU-STEM), confirming the compression threshold generalizes across languages and difficulties. Second, we conduct the first per-token perplexity analysis (n=723 tokens), revealing a "perplexity paradox": code syntax tokens are preserved (high perplexity) while numerical values in math problems are pruned despite being task-critical (low perplexity). Signature injection recovers +34 percentage points in pass rate (5.3% to 39.3%; Cohen's h=0.890). Third, we propose TAAC (Task-Aware Adaptive Compression), achieving 22% cost reduction with 96% quality preservation, outperforming fixed-ratio compression by 7%. MBPP validation (n=1,800 trials) confirms systematic variation: 3.6% at r=0.3 to 54.6% at r=1.0.
Beyond Perplexity: Multi-dimensional Safety Evaluation of LLM Compression
Kahlert School of Computing, University of Utah(犹他大学卡勒特计算学院);Google DeepMind(谷歌DeepMind);Scientific Computing and Imaging Institute, University of Utah(犹他大学科学计算与成像研究所)
The Lottery LLM Hypothesis, Rethinking What Abilities Should LLM Compression Preserve?
The Hong Kong University of Science and Technology(香港科技大学);The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));National University of Singapore(新加坡国立大学)
Divergent Token Metrics: Measuring degradation to prune away LLM components -- and optimize quantization
Aleph Alpha;Technical University Darmstadt(达姆施塔特工业大学);Hessian Center for Artificial Intelligence(黑森人工智能中心);LMU Munich(慕尼黑大学);German Center for Artificial Intelligence(德国人工智能研究中心);Munich Center for Machine Learning(慕尼黑机器学习中心)
FlattenGPT: Depth Compression for Transformer with Layer Flattening
State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University(多媒体信息处理国家重点实验室,计算机科学学院,北京大学);Tsinghua University(清华大学)
Memory-Efficient LLM Pretraining via Minimalist Optimizer Design
Department of Electrical and Computer Engineering, University of Minnesota, USA(电气与计算机工程系,明尼苏达大学,美国);School of Mathematics and Statistics, University of Sydney, Australia(数学与统计学学院,悉尼大学,澳大利亚)
Thinking with Drafting: Optical Decompression via Logical Reconstruction
Shenyang Institute of Computing Technology, Chinese Academy of Sciences(中国科学院沈阳计算技术研究所);University of Chinese Academy of Sciences 3 ByteDance 4 Westlake University(中国科学院大学 3 字节跳动 4 西湖大学);Key Laboratory of Computing Power Network(计算功率网络重点实验室);Information Security, Ministry of Education, Shandong Computer Science Center (National Supercomputer Center in Jinan), Qilu University of Technology (Shandong Academy of Sciences)(信息安全,教育部,山东计算机科学中心(济南国家超级计算中心),齐鲁工业大学(山东省科学院))
Persona-aware Generative Model for Code-mixed Language
Indian Institute of Technology Delhi(印度德里理工学院);Indraprastha Institute of Information Technology Delhi(因德拉普拉斯信息技术学院德里分校);Yardi School of Artificial Intelligence(亚迪人工智能学院)
SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compression
University of Washington(华盛顿大学);HSE University(高等经济大学);Yandex(扬德克斯);Skoltech(斯科尔科沃科学技术研究院);IST Austria(奥地利科学技术研究所);ETH Zurich(苏黎世联邦理工学院);NeuralMagic(神经魔法公司)