Summary of Scidfm: a Large Language Model with Mixture-of-experts For Science, by Liangtai Sun et al.
SciDFM: A Large Language Model with Mixture-of-Experts for Science
by Liangtai Sun, Danyu Luo, Da Ma, Zihan Zhao, Baocai Chen, Zhennan Shen, Su Zhu, Lu Chen, Xin Chen, Kai Yu
First submitted to arxiv on: 27 Sep 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 Recently, there has been a surge in using large language models (LLMs) to aid scientific discovery. However, most LLMs focus on general science and lack domain-specific knowledge, such as chemical molecules and amino acid sequences. To bridge this gap, we introduce SciDFM, a mixture-of-experts LLM trained from scratch, capable of conducting college-level scientific reasoning and understanding molecules and amino acid sequences. We collect a large-scale training corpus containing numerous scientific papers, books, and domain-specific databases. The pre-trained model is fine-tuned on instruction data to improve performance on downstream benchmarks. Our experiment results show that SciDFM achieves strong performance on general scientific benchmarks (SciEval, SciQ) and reaches state-of-the-art (SOTA) performance on domain-specific benchmarks among models of similar size. We analyze the expert layers and find varying results depending on data from different disciplines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine a super smart computer that can read and understand scientific books and articles. This computer is called SciDFM, and it’s really good at doing science problems. It was trained to learn about specific things like molecules and amino acid sequences. We made SciDFM by combining many smaller models that are experts in different areas of science. We tested SciDFM on many problems and it did very well! It’s so good that it even beats other computers that are similar in size. This is important because it means we can use SciDFM to help scientists do their work better. |
Keywords
» Artificial intelligence » Mixture of experts
