ChatGLM-6B is an open bilingual language model based on [General Language Model (GLM)](https://github.com/THUDM/GLM) framework, with 6.2 billion parameters. With the quantization technique, users can deploy locally on consumer-grade graphics cards (only 6GB of GPU memory is required at the INT4 quantization level).
ChatGLM-6B uses technology similar to ChatGPT, optimized for Chinese QA and dialogue. The model is trained for about 1T tokens of Chinese and English corpus, supplemented by supervised fine-tuning, feedback bootstrap, and reinforcement learning wit human feedback. With only about 6.2 billion parameters, the model is able to generate answers that are in line with human preference.
**[2023/03/31]** Added a parameter-efficient tuning implementation based on [P-Tuning-v2](https://github.com/THUDM/P-tuning-v2). The minimum INT4 quantization level only needs 7GB GPU memory is enough for model tuning. See [Parameter-efficient tuning method](ptuning/README.md) for details.
**[2023/03/23]** Add API deployment, thanks to [@LemonQu-GIT](https://github.com/LemonQu-GIT). Add embedding-quantized model [ChatGLM-6B-INT4-QE](https://huggingface.co/THUDM/chatglm-6b-int4-qe). Add support for GPU inference on Mac with Apple Silicon.
**[2023/03/19]** Add streaming output function `stream_chat`, already applied in web and CLI demo. Fix Chinese punctuations in output. Add quantized model [ChatGLM-6B-INT4](https://huggingface.co/THUDM/chatglm-6b-int4).
* [ChatGLM-MNN](https://github.com/wangzhaode/ChatGLM-MNN): An [MNN](https://github.com/alibaba/MNN)-based implementation of ChatGLM-6B C++ inference, which supports automatic allocation of computing tasks to GPU and CPU according to the size of GPU memory
* [ChatGLM-Tuning](https://github.com/mymusise/ChatGLM-Tuning): Fine-tuning ChatGLM-6B based on LoRA
If you have other good projects, please refer to the above format to add to README and propose [PR](https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/proposing-changes-to-your-work-with-pull-requests/creating-a-pull-request-from-a-fork).
Install the requirements with pip: `pip install -r requirements.txt`. `transformers` library version is recommended to be `4.27.1`, but theoretically any version no lower than `4.23.1` is acceptable.
In addition, if you need to run the quantified model on the CPU, you also need to install `gcc` and `openmp`. Most Linux distributions are installed by default. For Windows, you can check `openmp` when installing [TDM-GCC](https://jmeubank.github.io/tdm-gcc/). On Windows testing environment, the `gcc` version is `TDM-GCC 10.3.0`, and on Linux is `gcc 11.3.0`.
The command runs an interactive program in the shell. Type your instruction in the shell and hit enter to generate the response. Type `clear` to clear the dialogue history and `stop` to terminate the program.
By default, the model parameters are loaded with FP16 precision, which require about 13GB of GPU memory. It your GPU memory is limited, you can try to load the model parameters with quantization:
After 2 to 3 rounds of dialogue, the GPU memory usage is about 10GB under 8-bit quantization, and only 6GB under 4-bit quantization. As the number of dialogue rounds increases, the corresponding GPU memory consumption also increases. Due to the use of relative position encoding, ChatGLM-6B theoretically supports an infinitely long context-length, but the performance will gradually decline after the total length exceeds 2048 (training length).
Model quantization brings a certain performance decline. After testing, ChatGLM-6B can still perform natural and smooth generation under 4-bit quantization. using [GPT-Q](https://arxiv.org/abs/2210.17323) etc. The quantization scheme can further compress the quantization accuracy/improve the model performance under the same quantization accuracy. You are welcome to submit corresponding Pull Requests.
**[2023/03/19]** The quantization costs about 13GB of CPU memory to load the FP16 model. If your CPU memory is limited, you can directly load the quantized model, which costs only 5.2GB CPU memory:
```python
model = AutoModel.from_pretrained("THUDM/chatglm-6b-int4", trust_remote_code=True).half().cuda()
If your encounter the error `Could not find module 'nvcuda.dll'` or `RuntimeError: Unknown platform: darwin`(MacOS), please refer to this [Issue](https://github.com/THUDM/ChatGLM-6B/issues/6#issuecomment-1470060041).
### GPU Inference on Mac
For Macs (and MacBooks) with Apple Silicon, it is possible to use the MPS backend to run ChatGLM-6B on the GPU. First, you need to refer to Apple's [official instructions](https://developer.apple.com/metal/pytorch) to install PyTorch-Nightly. Then clone the model repository locally (you need to [install Git LFS](https://docs.github.com/zh/repositories/working-with-files/managing-large-files/installing-git-large-file-storage)）
```shell
git lfs install
git clone https://huggingface.co/THUDM/chatglm-6b
```
Change the code to load the model from your local path, and use the mps backend:
```python
model = AutoModel.from_pretrained("your local path", trust_remote_code=True).half().to('mps')
```
Then you can use GPU-accelerated model inference on Mac.
Parameter-efficient tuning based on [P-tuning v2](https://github.com/THUDM/P-tuning-v2). See [ptuning/README.md](ptuning/README.md) for details on how to use it.
This repository is licensed under the [Apache-2.0 License](LICENSE). The use of ChatGLM-6B model weights is subject to the [Model License](MODEL_LICENSE)。
## Citation
If you find our work useful, please consider citing the following papers:
```
@inproceedings{
zeng2023glm-130b,
title={{GLM}-130B: An Open Bilingual Pre-trained Model},
author={Aohan Zeng and Xiao Liu and Zhengxiao Du and Zihan Wang and Hanyu Lai and Ming Ding and Zhuoyi Yang and Yifan Xu and Wendi Zheng and Xiao Xia and Weng Lam Tam and Zixuan Ma and Yufei Xue and Jidong Zhai and Wenguang Chen and Zhiyuan Liu and Peng Zhang and Yuxiao Dong and Jie Tang},
booktitle={The Eleventh International Conference on Learning Representations (ICLR)},
