2 papers
cs.AI2026
ChartAnno: Evaluating MLLMs for Chart Annotation Generation
Zhenghan Chen, Zekai Shao, Lidan Tan +10
Multimodal large language models (MLLMs) have made significant progress in chart understanding, generation, and editing, but their ability to annotate existing charts remains under…
cs.CL2026
Does Accuracy Equal Evidence? Reasoning Faithfulness under KV Cache Compression
Mengting Ai, Jingrui He, Yue Guo
KV cache compression is commonly evaluated by final-answer accuracy, implicitly assuming that preserving the answer also preserves the reasoning that supports it. We test this assu…