Physics of Language Models: Part 3.2, Knowledge Manipulation

arxiv:2309.14402

Original source · Paper JSON

Reasoning and the "physics" of language models · 2023-09-25

Cites (6) · Cited by (10)

Complete cleaned Markdown · Source Markdown including bibliography

Read in parts

1. Part 1
2. Part 2
3. Part 3
4. Part 4
5. Part 5

Contents

• Physics of Language Models: Part 3.2, Knowledge Manipulation Thanks: Project page: https://physics.allen-zhu.com/part-3-knowledge/part-3-2 . An extended video of this paper is available at https://youtu.be/YSHzKmEianc . V1 was circulated internally at Meta on Sep 18, 2023, and appeared on arXiv on Sep 25, 2023. V2 polishes writing and includes additional Llama/Mistral experiments and larger data; but the conclusions remain unchanged. We would like to thank Lin Xiao, Chunting Zhou, Xiaodong Liu, Zhijie Zhou for many helpful conversations. We would like to extend special thanks to Nabib Ahmed, Giri Anantharaman, Lucca Bertoncini, Henry Estela, Liao Hu, Caleb Ho, Wil Johnson, Apostolos Kokolis, and Shubho Sengupta from Meta FAIR, as well as Ian Clark, Gourab De, Anmol Mann, and Max Pfeifer from W&B ; without their invaluable support, the experiments in this paper would not have been possible.
• Abstract
• 1 Introduction
• 1.1 Our Results
• Remark 1.1 .
• 1.2 Our Contributions
• 2 Preliminaries
• 3 Results 1-2: Knowledge Dual and Partial Retrievals
• 4 Results 3-6: Knowledge Classification and Comparison
• 5 Results 7-9: Knowledge Inverse Search
• 6 Conclusion
• Appendix A More Details on Data Preparation
• A.1 BIO dataset bioS
• Remark A.1 .
• A.1.1 Adding Reverse Knowledge
• A.2 BIO dataset bioR
• Appendix B More Details on Knowledge Retrieval
• Remark B.1 .
• Appendix C More Details on Knowledge Classification and Comparison
• Appendix D More Details on Knowledge Inverse Search
• Appendix E More Details on ChatGPT Experiments
• E.1 Inverse Knowledge Search
• E.2 Knowledge Classification and Comparison