Reverse Training to Nurse the Reversal Curse
This addresses a fundamental limitation in LLMs for AI researchers and practitioners, though it is an incremental improvement over existing training methods.
The paper tackles the Reversal Curse in large language models, where they fail to generalize from 'A has B' to 'B is of A', by proposing reverse training that doubles tokens by using words twice in forward and reverse directions, showing it improves performance on standard tasks and resolves reversal issues.
Large language models (LLMs) have a surprising failure: when trained on "A has a feature B", they do not generalize to "B is a feature of A", which is termed the Reversal Curse. Even when training with trillions of tokens this issue still appears due to Zipf's law - hence even if we train on the entire internet. This work proposes an alternative training scheme, called reverse training, whereby all words are used twice, doubling the amount of available tokens. The LLM is trained in both forward and reverse directions by reversing the training strings while preserving (i.e., not reversing) chosen substrings, such as entities. We show that data-matched reverse-trained models provide superior performance to standard models on standard tasks, and compute-matched reverse-trained models provide far superior performance on reversal tasks, helping resolve the reversal curse issue.