The authors introduce ACE-Compression, a framework for measuring how post-training compression changes four deployment-facing capability areas: procedural reasoning and constraint execution, knowledge grounding and truthfulness, social cognition and affective understanding, and persona fidelity and stylistic consistency.
Compression preserves broad LLM skills while weakening context-sensitive abilities
Across six open-weight model families, ACE-Compression finds that quantization retains underlying capabilities more reliably than pruning.
Top University
Jiahui Ning · Liang Chen · Yuchen Liang · Yi Zhang · Fenglei Fan
City University of Hong Kong · Tongji University · Rensselaer Polytechnic Institute · Sichuan University
Research Digest··3 min read
Ning and colleagues evaluate more than 35 compressed checkpoints using ten benchmarks designed to probe four broad ability dimensions, rather than relying only on aggregate task accuracy.
§