Summary of Autoft: Learning An Objective For Robust Fine-tuning, by Caroline Choi et al.
AutoFT: Learning an Objective for Robust Fine-Tuning
by Caroline Choi, Yoonho Lee, Annie Chen, Allan Zhou, Aditi Raghunathan, Chelsea Finn
First submitted to arxiv on: 18 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes AutoFT, a data-driven approach for robust fine-tuning of foundation models. Foundation models are pre-trained on large datasets and can be adapted to specific tasks through fine-tuning. However, current methods for fine-tuning often result in decreased performance under distribution shifts. To address this issue, AutoFT uses bi-level optimization to search for an objective function and hyperparameters that maximize out-of-distribution generalization. The approach is evaluated on nine natural distribution shifts, demonstrating significant improvement over existing methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary AutoFT is a new way to fine-tune foundation models so they work well even when the data is different from what they were trained on. This is important because real-world datasets can be quite different from the ones used during training. The approach uses a special kind of optimization called bi-level optimization, which helps find the best way to adapt the model for a specific task. The results show that AutoFT does much better than other methods in this area. |
Keywords
* Artificial intelligence * Fine tuning * Generalization * Objective function * Optimization
