Similar Items: Efficient Pre-Training with Token Superposition
- PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization
- Long Context Pre-Training with Lighthouse Attention
- Cubit: Token Mixer with Kernel Ridge Regression
- The First Token Knows: Single-Decode Confidence for Hallucination Detection
- Fuzzy Fingerprinting Encoder Pre-trained Language Models for Emotion Recognition in Conversations: Human Assessment and Validity Study
- A Comprehensive Analysis of Tokenization and Self-Supervised Learning in End-to-End Automatic Speech Recognition applied on French Language