Tongjilibo/bert4torch

Health v2 · maintenance only

Adoption not part of the score

1328 stars · 166 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

bert4torch is a PyTorch library providing an elegant reimplementation of transformer models (BERT, RoBERTa, T5, GPT, ChatGLM, LLaMA, etc.) for inference, fine-tuning, and building custom NLP models. It includes rich examples for tasks like text classification, sequence labeling, and LLM deployment via a one-line command-line service.

Use cases

• fine-tune bert models for text classification in pytorch
• run inference with chatglm or llama weights
• train a named entity recognition model
• deploy a large language model service from the command line
• load pretrained transformer weights and customize the model architecture
• train a seq2seq model for text generation

When to choose

• you want a lightweight, readable PyTorch alternative to HuggingFace transformers for BERT-style models
• you need to fine-tune Chinese LLMs like ChatGLM, Baichuan, or BELLE
• you prefer bert4keras-style APIs but work in PyTorch

When to avoid

• you need the broadest model hub coverage and ecosystem integrations of HuggingFace transformers
• you require production-grade multi-GPU serving at scale
• your team depends on TensorFlow rather than PyTorch

Facets

library · maturity active

machine-learning llm-inference llm-training nlp transformers large-language-models machine-learning deep-learning python bert pytorch transformers fine-tuning named-entity-recognition text-classification seq2seq chatglm llama pretrained-models natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Tongjilibo/bert4torch")