---
license: apache-2.0
language:
- zh
- en
pipeline_tag: text-generation
library_name: transformers
---
GitHub Repo |
Technical Report
👋 Join us on Discord and WeChat
## Overview
BitCPM-CANN-3B-unquantized is the **unquantized QAT (Quantization-Aware Training) checkpoint** of BitCPM-CANN-3B, designed for **continued pre-training and fine-tuning**. It preserves full-precision latent weights with ternary fake quantizers (weights → {-1, 0, 1} with group-wise scaling, trained via STE) defined in `modeling.py`, enabling the model to keep learning under quantization constraints. For technical details, see our [Technical Report](https://github.com/OpenBMB/MiniCPM/blob/main/docs/BitCPM_CANN.pdf).
> ⚠️ **This model is NOT for direct inference.** For inference, use the pseudo-quantized version: [openbmb/BitCPM-CANN-3B](https://huggingface.co/openbmb/BitCPM-CANN-3B).
## Continued Pre-training & Fine-tuning
The **only requirement** is that the forward pass must go through the bundled `modeling.py` (which contains the ternary fake quantizer). Load with `trust_remote_code=True` and do NOT replace or bypass the model's forward logic.
### Option 1: DeepSpeed (Recommended)
We provide ready-to-use training scripts in the [example](https://huggingface.co/openbmb/BitCPM-CANN-3B-unquantized/tree/main/example) directory (using the 1B model as an example):
- **Continued pre-training**: `example/run.sh` + `example/train.py`
- **SFT (Supervised Fine-tuning)**: `example/run_sft.sh` + `example/train_sft.py`
Quick start:
```bash
# Continued pre-training
cd example && bash run.sh
# Supervised fine-tuning
cd example && bash run_sft.sh
```
### Option 2: HuggingFace-compatible Frameworks
Any framework that supports HuggingFace model loading with custom code can be used, such as **LLaMA Factory**, **HuggingFace Trainer**, etc. The key is to ensure `trust_remote_code=True`:
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
path = 'openbmb/BitCPM-CANN-3B-unquantized'
tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
path,
torch_dtype=torch.bfloat16,
trust_remote_code=True
)
# Use with your preferred framework (LLaMA Factory, HF Trainer, etc.)
# The ternary fake quantizer in modeling.py is applied automatically during forward pass.
```
## Post-Training Conversion
After training, use `qat-convert.py` to fuse the fake quantizer and produce inference-ready pseudo-quantized weights:
```bash
python qat-convert.py \
--input_bin \
--output \
--quant_type ternary \
--group_size -1
```
The converted model can be loaded for inference in the same way as [openbmb/BitCPM-CANN-3B](https://huggingface.co/openbmb/BitCPM-CANN-3B)—no special quantization libraries required.
## Workflow
```
┌─────────────────────────────────┐
│ BitCPM-CANN-3B-unquantized │ ← This model (QAT checkpoint + fake quantizer in modeling.py)
└───────────────┬─────────────────┘
│
▼ Train (DeepSpeed / LLaMA Factory / HF Trainer / ...)
┌─────────────────────────────────┐
│ Fine-tuned checkpoint │ ← Still contains un-fused QAT parameters
└───────────────┬─────────────────┘
│
▼ python qat-convert.py --quant_type ternary --group_size -1
┌─────────────────────────────────┐
│ Pseudo-quantized model │ ← Ready for inference (same format as BitCPM-CANN-3B)
└─────────────────────────────────┘
```
## BitCPM-CANN Model Family
| Model | HuggingFace (Inference) | HuggingFace (Fine-tuning) |
|-------|-------------------------|---------------------------|
| BitCPM-CANN-0.5B | [openbmb/BitCPM-CANN-0.5B](https://huggingface.co/openbmb/BitCPM-CANN-0.5B) | [openbmb/BitCPM-CANN-0.5B-unquantized](https://huggingface.co/openbmb/BitCPM-CANN-0.5B-unquantized) |
| BitCPM-CANN-1B | [openbmb/BitCPM-CANN-1B](https://huggingface.co/openbmb/BitCPM-CANN-1B) | [openbmb/BitCPM-CANN-1B-unquantized](https://huggingface.co/openbmb/BitCPM-CANN-1B-unquantized) |
| BitCPM-CANN-3B | [openbmb/BitCPM-CANN-3B](https://huggingface.co/openbmb/BitCPM-CANN-3B) | [openbmb/BitCPM-CANN-3B-unquantized](https://huggingface.co/openbmb/BitCPM-CANN-3B-unquantized) |
| BitCPM-CANN-8B | [openbmb/BitCPM-CANN-8B](https://huggingface.co/openbmb/BitCPM-CANN-8B) | [openbmb/BitCPM-CANN-8B-unquantized](https://huggingface.co/openbmb/BitCPM-CANN-8B-unquantized) |
## Statement
- As a language model, BitCPM-CANN generates content by learning from a vast amount of text.
- However, it does not possess the ability to comprehend or express personal opinions or value judgments.
- Any content generated by BitCPM-CANN does not represent the viewpoints or positions of the model developers.
- Therefore, when using content generated by BitCPM-CANN, users should take full responsibility for evaluating and verifying it on their own.
## LICENSE
- This repository and BitCPM-CANN models are released under the [Apache-2.0](https://github.com/OpenBMB/MiniCPM/blob/main/LICENSE) License.
## Citation
- Please cite our technical report if you find our work valuable.
```bibtex
@article{bitcpmcann,
title={{BitCPM-CANN}: Native 1.58-Bit Large Language Model Training on Ascend NPU},
author={BitCPM Team},
year={2026}
}
```
