Summary of Toward Understanding In-context Vs. In-weight Learning, by Bryan Chan et al.
Toward Understanding In-context vs. In-weight Learning
by Bryan Chan, Xinyi Chen, András György, Dale Schuurmans
First submitted to arxiv on: 30 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 The paper investigates the emergence and disappearance of in-context learning in transformers when certain distributional properties are present in the training data. Researchers identify simplified distributional properties that give rise to this phenomenon, analyzing a simplified model with a gating mechanism and generalization error/ regret analysis. The findings are corroborated experimentally using a full transformer on stylized distributions. The study is extended to a large language model, showing how fine-tuning can elicit similar in-context learning behavior. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at why transformers learn new things when shown examples of what they’re supposed to do. It finds some simple rules that make this happen and tests it with a simplified version of the transformer. The results match what happens with the real transformer model, and it shows how fine-tuning on different types of text can help the model learn even more. |
Keywords
» Artificial intelligence » Fine tuning » Generalization » Large language model » Transformer
