Test-Time Training on Nearest Neighbors for Large Language Models
This addresses the problem of high costs in retrieval-augmented language models for researchers and practitioners, though it is incremental as it builds on existing fine-tuning and retrieval methods.
The paper tackles the computational and memory inefficiencies of augmenting language models with retrieved data by fine-tuning the model on retrieved neighbors at test time, resulting in significant performance improvements across over 20 language modeling tasks, such as narrowing the gap between a small GPT-2 and a much larger GPT-Neo model.
Many recent efforts augment language models with retrieval, by adding retrieved data to the input context. For this approach to succeed, the retrieved data must be added at both training and test time. Moreover, as input length grows linearly with the size of retrieved data, cost in computation and memory grows quadratically for modern Transformers. To avoid these complications, we simply fine-tune the model on retrieved data at test time, using its standard training setup. We build a large-scale distributed index based on text embeddings of the Pile dataset. For each test input, our system retrieves its neighbors and fine-tunes the model on their text. Surprisingly, retrieving and training on as few as 20 neighbors, each for only one gradient iteration, drastically improves performance across more than 20 language modeling tasks in the Pile. For example, test-time training with nearest neighbors significantly narrows the performance gap between a small GPT-2 and a GPT-Neo model more than 10 times larger. Sufficient index quality and size, however, are necessary. Our work establishes a first baseline of test-time training for language modeling.