OLMo: Accelerating the Science of Language Models
Groeneveld et al., 2024, OLMo (Proceedings of the 62nd Annual Meeting of ACL (Volume 1: Long Papers))
Subscriber deck — free during early access
1 cards in “2.1 OLMo Model and Architecture” · whole deck
2.1 OLMo Model and Architecture
- What changes did OLMo make to the original Transformer?No bias terms (for stability); non-parametric layer norm without learned gain or bias (judged safest and fastest); SwiGLU activations (hidden size about \(\tfrac{8}{3}d\), rounded up to a multiple of 128); rotary position embeddings; and a BPE tokenizer with extra tokens for masking personal information.
- Groeneveld et al., 2024, OLMo (Proceedings of the 62nd Annual Meeting of ACL (Volume 1: Long Papers)), §2.1, p. 15790
Source
- Dirk Groeneveld, Iz Beltagy, Evan Walsh, Akshita Bhagia, Rodney Kinney, Oyvind Tafjord, Ananya Jha, Hamish Ivison, Ian Magnusson, Yizhong Wang, Shane Arora, David Atkinson, Russell Authur, Khyathi Chandu, Arman Cohan, Jennifer Dumas, Yanai Elazar, Yuling Gu, Jack Hessel, Tushar Khot, William Merrill, Jacob Morrison, Niklas Muennighoff, Aakanksha Naik, Crystal Nam, Matthew Peters, Valentina Pyatkin, Abhilasha Ravichander, Dustin Schwenk, Saurabh Shah, William Smith, Emma Strubell, Nishant Subramani, Mitchell Wortsman, Pradeep Dasigi, Nathan Lambert, Kyle Richardson, Luke Zettlemoyer, Jesse Dodge, Kyle Lo, Luca Soldaini, Noah Smith and Hannaneh Hajishirzi. 2024. OLMo: Accelerating the Science of Language Models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 15789–15809. License: CC BY 4.0. Flashcards adapted from this work. doi:10.18653/v1/2024.acl-long.841 https://aclanthology.org/2024.acl-long.841/