---
license: mit
language: en
tags:
- LLM
- baichuan
---
## Special Notes

We have released **7 lyraBaichuan models** including lyraBaichuan-7B, lyraBaichuan-13B-Base, lyraBaichuan-13B-Chat, lyraBaichuan2-7B-Base, lyraBaichuan2-7B-Chat, lyraBaichuan2-13B-Base and lyraBaichuan2-13B-Chat. 

These highly optimized Baichuan models are suitable for Ampere (A100/A10) as well as Volta (V100).

If you like our work and consider to join us, feel free to drop a line at benbinwu@tencent.com.

## Model Card for lyraBaichuan

lyraBaichuan is currently the **fastest Baichuan models** (Baichuan-7B, Baichuan-13B, Baichuan2-7B, Baichuan2-13B) available. The inference speed of lyraBaichuan has achieved up to **4300+ tokens/s** on A100, up to **2.4x** acceleration upon the torch version.

Among its main features are:
- device: Nvidia GPU with Amperer architecture or Volta architecture (A100 or higher, V100).
- batch_size: compiled with dynamic batch size, maximum depends on device. 
- MEMOPT mode: significantly optimized VRAM usage and increased speed

We use the Baichuan2-7B-Base and Baichuan2-13B-Base model for measurement, but this optimized inference is also applicable to other Baichuan models, including Baichuan-7B and Baichuan-13B.

## Speed

* Evaluated at tokens/s (#tokens of input and output divided by inference time cost)
* test on A100 40G
* MEMOPT mode

### Baichuan2-7B-Base

| Version | Batch Size 1 | Batch Size 8 | Batch Size 16 | Batch Size 32 | Batch Size 64 |
| --- | --- | --- | --- | --- | --- |
| Torch 2.0.1 | 41.2 | 323.2 | 640.0 | 1256.8 | 2231.0 |
| lyraBaichuan | 125.9 | 948.1 | 1749.3 | 2974.0 | 4370.1 |

### Baichuan2-13B-Base

| Version | Batch Size 1 | Batch Size 8 | Batch Size 16 | Batch Size 32 | Batch Size 64 |
| --- | --- | --- | --- | --- | --- |
| Torch 2.0.1 | 40.9 | 307.9 | 555.6 | 1010.4 | 1601.0 |
| lyraBaichuan | 80.0 | 568.2 | 1124.4 | 1942.6 | 2828.0 |

## Docker Environment Recommendation

- For Cuda 11.X: we recommend ```nvcr.io/nvidia/pytorch:22.12-py3```
- For Cuda 12.0: we recommend ```nvcr.io/nvidia/pytorch:23.02-py3```

```bash
docker pull nvcr.io/nvidia/pytorch:23.02-py3
docker run --rm -it --gpus all -v ./:/lyraBaichuan nvcr.io/nvidia/pytorch:23.02-py3

pip install -r requirements.txt
python demo.py
```

## Uses

```python
from lyra_baichuan import lyraBaichuan7B, lyraBaichuan13B

model_path = "./models/Baichuan2-13B-lyra"
tokenizer_path = "./models/Baichuan2-13B-lyra"
inference_dtype = 'fp16'
prompt = "登鹳雀楼->王之涣\n夜雨寄北->"

memopt_mode = 1
max_output_length = 64
arch = "Ampere" # Ampere or Volta
cuda_version = 12 # cuda version, we currently support 11 and 12

# To use 7B model, initialize with lyraBaichuan7B
model = lyraBaichuan13B(model_path, 
                        tokenizer_path = tokenizer_path, 
                        dtype = inference_dtype,
                        memopt_mode = memopt_mode,
                        arch = arch,
                        cuda_version = cuda_version)

bs = 1
prompts = [prompt, ] * bs
output_texts = model.generate(
        prompts, output_length=max_output_length,
        top_k=30, top_p=0.85, temperature=1.0, repetition_penalty=1.0, do_sample=False)

print(output_texts)
```

## Demo Outputs

### Baichuan2-13B-Base
#### input

登鹳雀楼->王之涣

夜雨寄北->

#### output

李商隐

望洞庭->刘禹锡

黄鹤楼送孟浩然之广陵->李白

登岳阳楼->杜甫

秋词->刘禹锡

枫桥夜泊->张继

饮湖上初晴后雨->苏轼

浪淘沙->刘禹锡

## TODO
1. Support for int4
2. Inference for longer context situations
3. Streaming inference mode.

## Citation
``` bibtex
@Misc{lyraBaichuan2023,
  author =       {Haoxiong Su, Kangjian Wu, Zhengtao Wang, Yibo Lu, Bin Wu},
  title =        {lyraBaichuan: Accelerating Baichuan models to 4300+ tokens/s},
  howpublished = {\url{https://huggingface.co/TMElyralab/lyraBaichuan}},
  year =         {2023}
}
```

## Report bug
- start a discussion to report any bugs!--> https://huggingface.co/TMElyralab/lyraBaichuan
- report bug with a `[bug]` mark in the title.