Merge pull request #62 from GanymedeNil/main

Add some parameter support in web demo
pull/157/head
Zhengxiao Du 2023-03-17 17:37:07 +08:00 committed by GitHub
commit ecd2857c1a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
11 changed files with 135 additions and 8 deletions

63
.github/ISSUE_TEMPLATE/bug_report.yaml vendored Normal file
View File

@ -0,0 +1,63 @@
name: 🐞 Bug/Help
description: File a bug/issue
title: "[BUG/Help] <title>"
labels: []
body:
- type: checkboxes
attributes:
label: Is there an existing issue for this?
description: Please search to see if an issue already exists for the bug you encountered.
options:
- label: I have searched the existing issues
required: true
- type: textarea
attributes:
label: Current Behavior
description: |
A concise description of what you're experiencing, with screenshot attached if possible.
Tip: You can attach images or log files by clicking this area to highlight it and then dragging files in.
validations:
required: true
- type: textarea
attributes:
label: Expected Behavior
description: A concise description of what you expected to happen.
validations:
required: false
- type: textarea
attributes:
label: Steps To Reproduce
description: Steps to reproduce the behavior.
placeholder: |
1. In this environment...
2. With this config...
3. Run '...'
4. See error...
validations:
required: true
- type: textarea
attributes:
label: Environment
description: |
examples:
- **OS**: Ubuntu 20.04
- **Python**: 3.8
- **Transformers**: 4.26.1
- **PyTorch**: 1.12
- **CUDA Support**: True
value: |
- OS:
- Python:
- Transformers:
- PyTorch:
- CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) :
render: markdown
validations:
required: true
- type: textarea
attributes:
label: Anything else?
description: |
Links? References? Anything that will give us more context about the issue you are encountering!
validations:
required: false

1
.github/ISSUE_TEMPLATE/config.yml vendored Normal file
View File

@ -0,0 +1 @@
blank_issues_enabled: false

View File

@ -0,0 +1,26 @@
name: Feature request
description: Suggest an idea for this project
title: "[Feature] <title>"
labels: []
body:
- type: textarea
attributes:
label: Is your feature request related to a problem? Please describe.
description: |
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
validations:
required: false
- type: textarea
attributes:
label: Solutions
description: |
Describe the solution you'd like
A clear and concise description of what you want to happen.
validations:
required: true
- type: textarea
attributes:
label: Additional context
description: Add any other context or screenshots about the feature request here.
validations:
required: false

View File

@ -2,9 +2,10 @@
## 介绍
ChatGLM-6B 是一个开源的、支持中英双语的对话语言模型,基于 [General Language Model (GLM)](https://github.com/THUDM/GLM) 架构,具有 62 亿参数。结合模型量化技术用户可以在消费级的显卡上进行本地部署INT4 量化级别下最低只需 6GB 显存。ChatGLM-6B 使用了和 ChatGPT 相似的技术,针对中文问答和对话进行了优化。经过约 1T 标识符的中英双语训练辅以监督微调、反馈自助、人类反馈强化学习等技术的加持62 亿参数的 ChatGLM-6B 已经能生成相当符合人类偏好的回答。更多信息请参考我们的[博客](https://chatglm.cn/blog)。
ChatGLM-6B 是一个开源的、支持中英双语的对话语言模型,基于 [General Language Model (GLM)](https://github.com/THUDM/GLM) 架构,具有 62 亿参数。结合模型量化技术用户可以在消费级的显卡上进行本地部署INT4 量化级别下最低只需 6GB 显存)。
ChatGLM-6B 使用了和 ChatGPT 相似的技术,针对中文问答和对话进行了优化。经过约 1T 标识符的中英双语训练辅以监督微调、反馈自助、人类反馈强化学习等技术的加持62 亿参数的 ChatGLM-6B 已经能生成相当符合人类偏好的回答。更多信息请参考我们的[博客](https://chatglm.cn/blog)。
同时,我们基于千亿基座的[ChatGLM 模型](https://chatglm.cn)正在邀请制内测,后续将逐步扩大内测范围,欢迎申请加入内测
不过由于ChatGLM-6B的规模较小目前已知其具有相当多的[**局限性**](#局限性),如事实性/数学逻辑错误,可能生成有害/有偏见内容,较弱的上下文能力,自我认知混乱,以及对英文指示生成与中文指示完全矛盾的内容。请大家在使用前了解这些问题,以免产生误解
*Read this in [English](README_en.md).*
@ -46,7 +47,7 @@ ChatGLM-6B 是一个开源的、支持中英双语的对话语言模型,基于
如果这些方法无法帮助你入睡,你可以考虑咨询医生或睡眠专家,寻求进一步的建议。
```
完整的模型实现可以在 [Hugging Face Hub](https://huggingface.co/THUDM/chatglm-6b) 上查看。
完整的模型实现可以在 [Hugging Face Hub](https://huggingface.co/THUDM/chatglm-6b) 上查看。如果你从Hugging Face Hub上下载checkpoint的速度较慢也可以从[这里](https://cloud.tsinghua.edu.cn/d/fb9f16d6dc8f482596c2/)手动下载。
### Demo
@ -107,6 +108,8 @@ model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).bf
```
需保证空闲内存接近16G并且推理速度会很慢。
MacOS 如果报错`RuntimeError: Unknown platform: darwin`的话请参考这个[Issue](https://github.com/THUDM/ChatGLM-6B/issues/6#issuecomment-1470060041).
## ChatGLM-6B示例
以下是一些使用`web_demo.py`得到的示例截图。更多ChatGLM-6B的可能等待你来探索发现
@ -163,6 +166,34 @@ model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).bf
</details>
## 局限性
由于ChatGLM-6B的小规模其能力仍然有许多局限性。以下是我们目前发现的一些问题
- 模型容量较小6B的小容量决定了其相对较弱的模型记忆和语言能力。在面对许多事实性知识任务时ChatGLM-6B可能会生成不正确的信息它也不擅长逻辑类问题如数学、编程的解答。
<details><summary><b>点击查看例子</b></summary>
![](limitations/factual_error.png)
![](limitations/math_error.png)
</details>
- 产生有害说明或有偏见的内容ChatGLM-6B只是一个初步与人类意图对齐的语言模型可能会生成有害、有偏见的内容。内容可能具有冒犯性此处不展示
- 英文能力不足ChatGLM-6B 训练时使用的指示/回答大部分都是中文的,仅有极小一部分英文内容。因此,如果输入英文指示,回复的质量远不如中文,甚至与中文指示下的内容矛盾,并且出现中英夹杂的情况。
- 易被误导对话能力较弱ChatGLM-6B 对话能力还比较弱,而且 “自我认知” 存在问题,并很容易被误导并产生错误的言论。例如当前版本的模型在被误导的情况下,会在自我认知上发生偏差。
<details><summary><b>点击查看例子</b></summary>
![](limitations/self-confusion_google.jpg)
![](limitations/self-confusion_openai.jpg)
![](limitations/self-confusion_tencent.jpg)
</details>
## 协议
本仓库的代码依照 [Apache-2.0](LICENSE) 协议开源ChatGLM-6B 模型的权重的使用则需要遵循 [Model License](MODEL_LICENSE)。

Binary file not shown.

After

Width:  |  Height:  |  Size: 134 KiB

BIN
limitations/math_error.png Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 25 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 152 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 143 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 125 KiB

View File

@ -2,3 +2,4 @@ protobuf>=3.19.5,<3.20.1
transformers>=4.26.1
icetk
cpm_kernels
torch>=1.10

View File

@ -9,10 +9,11 @@ MAX_TURNS = 20
MAX_BOXES = MAX_TURNS * 2
def predict(input, history=None):
def predict(input, max_length, top_p, temperature, history=None):
if history is None:
history = []
response, history = model.chat(tokenizer, input, history)
response, history = model.chat(tokenizer, input, history, max_length=max_length, top_p=top_p,
temperature=temperature)
updates = []
for query, response in history:
updates.append(gr.update(visible=True, value="用户:" + query))
@ -33,8 +34,12 @@ with gr.Blocks() as demo:
with gr.Row():
with gr.Column(scale=4):
txt = gr.Textbox(show_label=False, placeholder="Enter text and press enter").style(container=False)
txt = gr.Textbox(show_label=False, placeholder="Enter text and press enter", lines=11).style(
container=False)
with gr.Column(scale=1):
max_length = gr.Slider(0, 4096, value=2048, step=1.0, label="Maximum length", interactive=True)
top_p = gr.Slider(0, 1, value=0.7, step=0.01, label="Top P", interactive=True)
temperature = gr.Slider(0, 1, value=0.95, step=0.01, label="Temperature", interactive=True)
button = gr.Button("Generate")
button.click(predict, [txt, state], [state] + text_boxes)
demo.queue().launch(share=True)
button.click(predict, [txt, max_length, top_p, temperature, state], [state] + text_boxes)
demo.queue().launch(share=True, inbrowser=True)