---
license: apache-2.0
language:
- zh
base_model:
- hfl/chinese-roberta-wwm-ext
tags:
- finance
---

## Model Details

**Model Description:** This is a finance-domain pretrained Chinese language model, which is based on the 355-million-parameter RoBERTa-Large and further pre-trained on 32B tokens of Chinese financial corpora (including a large number of research reports, news, and announcements).

- **Developed by:** See [valuesimplex](https://github.com/valuesimplex) for model developers
- **Model Type:** Transformer-based language model
- **Language(s):** Chinese
- **Parent Model:** See the [chinese-roberta](https://huggingface.co/hfl/chinese-roberta-wwm-ext)  for more information about the BERT base model.
- **Resources for more information:**
  - [Research Paper](https://dl.acm.org/doi/10.1145/3711896.3737219)
  - [GitHub Repo](https://github.com/valuesimplex/FinBERT)

## Direct Use

```python
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("valuesimplex-ai-lab/FinBERT2-large")
tokenizer = AutoTokenizer.from_pretrained("valuesimplex-ai-lab/FinBERT2-large")
```

### Further Usage
continual pre-training or fine-tuning：https://github.com/valuesimplex/FinBERT