IDEA-CCNL/Fengshenbang-LM

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

4123 stars · 374 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Fengshenbang-LM is an open-source suite of Chinese large language models and a PyTorch training framework from IDEA Research's CCNL lab, aiming to be infrastructure for Chinese AIGC and cognitive intelligence. It includes pretrained models (Erlangshen, Ziya, Toulun), multimodal and domain models, and tools for distributed pretraining and finetuning.

Use cases

• download and finetune Chinese pretrained language models
• pretrain large language models with distributed training
• run Chinese text generation and AIGC tasks
• zero-shot text classification for Chinese NLP
• train multimodal models for Chinese named entity recognition
• generate Chinese images with a Chinese stable diffusion model

When to choose

• you need strong Chinese-language pretrained models or Chinese AIGC capabilities
• you want an open Apache-2.0 model suite with PyTorch and HuggingFace integration
• you need distributed pretraining/finetuning tooling for Chinese LLMs

When to avoid

• your work is purely English-centric with no Chinese NLP needs
• you only want inference serving rather than model training and research
• you need a lightweight plug-and-play library without GPU resources

Facets

library · maturity active

llm-training llm-inference machine-learning nlp transformers stable-diffusion large-language-models machine-learning deep-learning artificial-intelligence python windows chinese-nlp pretrained-models distributed-training aigc multimodal pytorch huggingface natural-language-processing linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="IDEA-CCNL/Fengshenbang-LM")