pretrained language models
These are language models that have been trained on large datasets prior to being fine-tuned for specific tasks. They leverage transfer learning to efficiently adapt to various applications like sentiment analysis or summarization.
- Brain-tuning Improves Generalizability and Efficiency of Brain Alignment in Speech Models
- Compositional Discrete Latent Code for High Fidelity, Productive Diffusion Models
- Emergence of Linear Truth Encodings in Language Models
- Extrapolation by Association: Length Generalization Transfer In Transformers
- Multi-Token Prediction Needs Registers
- Understanding Differential Transformer Unchains Pretrained Self-Attentions
- Words That Unite The World: A Unified Framework for Deciphering Central Bank Communications