Tag
The paper proposes DualMLC, a dual-branch framework that combines autoregressive and bidirectional language models for large-scale multi-label text classification, achieving state-of-the-art results on benchmarks.