CausalLM is not optimal for in-context learning.
Nan DingTomer LevinboimJialin WuSebastian GoodmanRadu SoricutPublished in: CoRR (2023)
Keyphrases
- learning algorithm
- learning process
- learning scheme
- unsupervised learning
- worst case
- learning tasks
- learning systems
- optimal solution
- context sensitive
- contextual information
- online learning
- domain knowledge
- active learning
- reinforcement learning
- supervised learning
- knowledge acquisition
- prior knowledge
- case study
- search algorithm
- learning scenarios
- context dependent
- machine learning