78 repos across 2 sub-areas
Pre-trained transformer-based language models, with a strong focus on BERT variants and multilingual/Chinese-language implementations. The cluster centers on model checkpoints, fine-tuning frameworks, and downstream NLP tasks enabled by transformers—particularly masked language modeling and encoder-based architectures. Repositories here span PyTorch and TensorFlow implementations, model weights, and tools for deploying and adapting these models to specific languages and domains.
BERT and Transformer Models
40 repos
Pre-trained language models based on the BERT architecture and transformer framework, covering both general-purpose and domain-specific variants (biomedical, multilingual, etc.). These repositories contain model weights, fine-tuning implementations, and inference endpoints built on PyTorch and TensorFlow. Developers exploring this cluster will find ready-to-use models for natural language understanding tasks like masked language modeling, text classification, and semantic similarity.
Multilingual BERT Models and Transformers
38 repos
Pre-trained BERT language models optimized for diverse languages and language pairs, including specialized variants for Russian, Italian, Swedish, Turkish, and multilingual combinations. These repositories provide transformer-based NLP foundations compatible with PyTorch and TensorFlow, enabling downstream tasks like text classification, named entity recognition, and semantic similarity across languages. The cluster represents the infrastructure layer for applying transfer learning and fine-tuning BERT architectures to non-English and cross-lingual NLP problems.