Linear Interpolation In Parameter Space is Good Enough for Fine-Tuned Language Models
This work provides a simple, efficient method for controllable text generation in NLP, though it is incremental as it builds on known interpolation techniques applied to fine-tuned models.
The paper tackled the problem of performance degradation during parameter interpolation in fine-tuned language models, finding that linear interpolation between fine-tuned parameters does not cause a performance drop in intermediate points. This result enables controllable text generation by adjusting attributes like sentiment without inference speed overhead.
The simplest way to obtain continuous interpolation between two points in high dimensional space is to draw a line between them. While previous works focused on the general connectivity between model parameters, we explored linear interpolation for parameters of pre-trained models after fine-tuning. Surprisingly, we could perform linear interpolation without a performance drop in intermediate points for fine-tuned models. For controllable text generation, such interpolation could be seen as moving a model towards or against the desired text attribute (e.g., positive sentiment), which could be used as grounds for further methods for controllable text generation without inference speed overhead.