Janek Falkenstein
2024
From Language to Pixels: Task Recognition and Task Learning in LLMs
Janek Falkenstein
|
Carolin M. Schuster
|
Alexander H. Berger
|
Georg Groh
Proceedings of the 2nd GenBench Workshop on Generalisation (Benchmarking) in NLP
LLMs can perform unseen tasks by learning from a few in-context examples. How in-context learning works is still uncertain. We investigate the mechanisms of in-context learning on a challenging non-language task. The task requires the LLM to generate pixel matrices representing images of basic shapes. We introduce a framework to analyze if this task is solved by recognizing similar formats from the training data (task recognition) or by understanding the instructions and learning the skill de novo during inference (task learning). Our experiments demonstrate that LLMs generate meaningful pixel matrices with task recognition and fail to learn such tasks when encountering unfamiliar formats. Our findings offer insights into LLMs’ learning mechanisms and their generalization ability to guide future research on their seemingly human-like behavior.
Search