Symbol tuning improves in-context learning in language models

A key feature of human intelligence is that humans can learn to perform new tasks by reasoning using only a few examples. Scaling up language models has unlocked a range of new applications and paradigms in machine learning, including the ability to perform challenging reasoning tasks via in-context learning. Language models, however, are still sensitive to the way that prompts are given, indicating that they are …

applications context examples feature google google research human human intelligence humans in-context learning intelligence jerry language language models learn machine machine intelligence machine learning natural language processing reasoning research researcher scaling scaling up