Learning from context is harder than we thought

Tencent Hunyuan introduces CL-bench, a benchmark to evaluate whether language models can learn new knowledge from context rather than relying solely on parametric memory, covering four types of real-world context learning scenarios with a contamination-free design.