GitHub - DAMO-NLP-MT/PolyLM
GitHub - DAMO-NLP-MT/PolyLMhttps://github.com/DAMO-NLP-MT/PolyLM

GitHub - DAMO-NLP-MT/PolyLM

🌐 Unlock the power of multilingual AI with PolyLM by DAMO-NLP-MT on GitHub! 🌍🔝 This cutting-edge language model covers 18 languages, excelling in major non-English ones. Enhance your text generation tasks with PolyLM-13B, PolyLM-Multialpaca-13B, and PolyLM-Chat-13B. #AI #NLP #GitHub

  • PolyLM is a polyglot large language model covering 18 commonly spoken languages, aiming to overcome limitations in current LLM research.
  • It excels in major non-English languages like Spanish, Russian, Arabic, Japanese, Korean, Thai, Indonesian, and Chinese.
  • PolyLM complements existing models like LLaMA and BLOOM by enhancing multilingual instruction-following capabilities.
  • It demonstrates strong performance across various tasks compared to other multilingual LLMs.
  • PolyLM is available in different scales on Modelscope and Huggingface.
  • Sample scripts using PolyLM-13B, PolyLM-Multialpaca-13B, and PolyLM-Chat-13B for text generation are provided.
  • Researchers and developers can freely use PolyLM-1.7B, PolyLM-13B, PolyLM-Multialpaca-13B, and PolyLM-Chat-13B codes and model weights.
  • The model authors are Xiangpeng Wei, Haoran Wei, Huan Lin, Tianhao Li, and others, with a paper published in 2023.
  • The model is licensed with Bibtex citation available.
  • A web demo using Gradio is accessible, and deployment locally requires significant GPU memory.
  • PolyLM is a significant contribution to the advancement of multilingual language models in various research and application domains.