Thinking Tokens for Language Modeling
This addresses a specific reasoning bottleneck in language models, but it is incremental as it builds on existing methods with a novel token-based approach.
The paper tackles the problem of language models making mistakes in complex calculations by proposing 'thinking tokens' that allow the model to perform more calculations when encountering difficult problems, resulting in enhanced generalization capability.
How much is 56 times 37? Language models often make mistakes in these types of difficult calculations. This is usually explained by their inability to perform complex reasoning. Since language models rely on large training sets and great memorization capability, naturally they are not equipped to run complex calculations. However, one can argue that humans also cannot perform this calculation immediately and require a considerable amount of time to construct the solution. In order to enhance the generalization capability of language models, and as a parallel to human behavior, we propose to use special 'thinking tokens' which allow the model to perform much more calculations whenever a complex problem is encountered.