Rugpt3small Based On Gpt2

🚀 rugpt3small_based_on_gpt2

This project presents a Russian language model based on the GPT - 2 architecture. The model's design, pre - training, and evaluation details are provided in the preprint. It was pre - trained and fine - tuned under specific settings by the SberDevices team.

🚀 Quick Start

The model architecture design, pretraining, and evaluation are documented in our preprint: A Family of Pretrained Transformer Language Models for Russian.

The model was pretrained with sequence length 1024 using transformers by the SberDevices team on 80B tokens around 3 epochs. After that, the model was finetuned with the context size of 2048.

Total training time took around one week on 32 GPUs.

✨ Features

• Pretrained on Large - Scale Data: The model was pre - trained on 80B tokens for about 3 epochs, which helps it capture rich language knowledge.
• Long - Context Fine - Tuning: It was fine - tuned with a context size of 2048, enabling better performance in handling long - text tasks.

👥 Authors

• NLP core team RnD Telegram channel:
  • Dmitry Zmitrovich

📄 Cite us

@misc{zmitrovich2023family,
      title={A Family of Pretrained Transformer Language Models for Russian}, 
      author={Dmitry Zmitrovich and Alexander Abramov and Andrey Kalmykov and Maria Tikhonova and Ekaterina Taktasheva and Danil Astafurov and Mark Baushenko and Artem Snegirev and Tatiana Shavrina and Sergey Markov and Vladislav Mikhailov and Alena Fenogenova},
      year={2023},
      eprint={2309.10931},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}