Galactica: A Large Language Model for Science (Drama & Paper Review)

#ai #galactica #meta Galactica is a language model trained on a curated corpus of scientific documents, such as papers, knowledge bases, reviews, and other articles. The model can be used in a generative fasion to assist scientific writing, do reference prediction, and much more, including a new approach to do step-by-step reasoning using a clever encoding of intermediate steps. This video explains the paper, but also dives into the drama that ensued once Meta released a public demo of the model. OUTLINE: 0:00 - Introduction 1:30 - Drama around the public demo 16:00 - Start of paper review 20:30 - Dataset construction and encoding 23:30 - Encoding step-by-step reasoning using a scratchpad 33:00 - Modelling scientific references & citations 35:05 - Prompt Pre-Training 37:10 - Architecture details 38:30 - Experimental results 49:20 - Conclusion Paper: https://galactica.org/static/paper.pdf Website: https://galactica.org/explore/ Abstract: Information overload is a major obstacle to scientific progress. The explosive growth in scientific literature and data has made it ever harder to discover useful insights in a large mass of information. Today scientific knowledge is accessed through search engines, but they are unable to organize scientific knowledge alone. In this paper we introduce Galactica: a large language model that can store, combine and reason about scientific knowledge. We train on a large scientific corpus of papers, reference material, knowledge bases and many other sources. We outperform existing models on a range of scientific tasks. On technical knowledge probes such as LaTeX equations, Galactica outperforms the latest GPT-3 by 68.2% versus 49.0%. Galactica also performs well on reasoning, outperforming Chinchilla on mathematical MMLU by 41.3% to 35.7%, and PaLM 540B on MATH with a score of 20.4% versus 8.8%. It also sets a new state-of-the-art on downstream tasks such as PubMedQA and MedMCQA dev of 77.6% and 52.9%. And despite not being trained on a general corpus, Galactica outperforms BLOOM and OPT-175B on BIG-bench. We believe these results demonstrate the potential for language models as a new interface for science. We…