Dear Colleagues,
please find below a beginner-friendly tutorial on In-Context Learning with Google Colab code:
In-context learning has recently been making waves in ML and robotics. The underlying idea is to use an LLM as the main engine for learning and adaptation. The LLM is provided with both the training data and the query and, in turn, generates the predicted label. We explain how you can solve fundamental ML and RL tasks using this approach. In particular, we provide demos and tutorials on how to solve tasks such as (a) optimization, (b) regression, (c) classification, (d) reinforcement learning, (e) self-improvement, (f) translation and many others using In-Context Learning.
All examples can be run using the free version of Gemini.
Best Regards,
Heni Ben Amor
Associate Professor
Arizona State University