Active Example Selection for In-Context Learning
This addresses the problem of selecting effective examples for in-context learning in language models, which is incremental as it builds on existing methods with specific gains.
The paper tackled the instability of in-context learning performance across different demonstration examples by formulating example selection as a sequential decision problem and using reinforcement learning to learn generalizable policies, resulting in a 5.8% average improvement on GPT-2 and a small improvement on GPT-3 Ada.
With a handful of demonstration examples, large-scale language models show strong capability to perform various tasks by in-context learning from these examples, without any fine-tuning. We demonstrate that in-context learning performance can be highly unstable across samples of examples, indicating the idiosyncrasies of how language models acquire information. We formulate example selection for in-context learning as a sequential decision problem, and propose a reinforcement learning algorithm for identifying generalizable policies to select demonstration examples. For GPT-2, our learned policies demonstrate strong abilities of generalizing to unseen tasks in training, with a $5.8\%$ improvement on average. Examples selected from our learned policies can even achieve a small improvement on GPT-3 Ada. However, the improvement diminishes on larger GPT-3 models, suggesting emerging capabilities of large language models.