题意:
oobabooga-textgen-web-ui how to get authorization to view model list from port 5000 via the ooba's api-key in python
怎样在Python中使用oobabooga的API密钥,通过端口5000获取模型列表的授权
问题背景:
I wish to extract and print out a list of llm models from the oobabooga-text-gen-web-ui in python.
我希望从Python中的oobabooga-text-gen-web-ui中提取并打印出一系列LLM(大型语言模型)的列表
Some context first before i head over to my problem.
在我描述我的问题之前,先提供一些背景信息。
For those familiar with what ooba is its essentially a Gradio web UI for Large Language Models.
对于那些熟悉ooba的人来说,它本质上是一个为大型语言模型(Large Language Models)设计的Gradio网络用户界面。
I downloaded and loaded a few llm models onto this web ui. The web ui uses http://127.0.0.1:7860/ to display the web user interface on port 7860. But if I enable the openai
and api
extensions, and also edit the CMD_FLAGS.txt within the ooba folder into something like:
我下载并在这个网络用户界面上加载了几个大型语言模型(LLM)。该网络用户界面使用 http://127.0.0.1:7860/
在7860端口上显示用户界面。但是,如果我启用了openai和api扩展,并且在ooba文件夹内编辑CMD_FLAGS.txt
文件,内容类似于:
python
--listen --api --api-key "enter-your-fake-api-key-here"
the extensions will mimic an Open AI api key by connecting to ooba from a network via port 5000
这些扩展将通过端口5000从网络连接到ooba,从而模拟一个OpenAI API密钥。
Here come the problem... 问题出现了
this is my code to view the model list from ooba : 以下是我用来查看ooba中模型列表的代码
python
import requests
url = "http://127.0.0.1:5000/v1"
#Model List
headers = {
"Content-Type": "application/json"
}
response = requests.get(f'{url}/internal/model/list',
headers=headers,
verify=False)
print(response.json())
the output should look something like this: 输出应该看起来像这样:
python
{'model_names': ['L3-8B-Stheno-v3.2', 'L3-8B-Stheno-v3.2_exl2_8h_8bpw', 'L3-8B-Stheno-v3.2_q8_0.gguf', 'MixTAO-7Bx2-MoE-v8.1_q8_0.gguf']}
but instead i got this: 但我得到了这个:
python
{'detail': 'Unauthorized'}
After some fiddling around I found that if I leave CMD_FLAGS.txt blank the code works as intended and i get the model lists but i dont have access to an API key since its not enable on ooba.
经过一些尝试后,我发现如果我将CMD_FLAGS.txt
文件留空,代码就能按预期工作,我可以获取到模型列表。但是,由于我没有在ooba上启用API密钥功能,因此我无法访问API密钥。
if I do enable it with CMD_FLAGS.txt and typing: 如果我在CMD_FLAGS.txt
中启用它并输入:
python
--listen --api --api-key "enter-your-fake-api-key-here"
I'll have access to the openai api key but the code to extract the model list will return as : {'detail': 'Unauthorized'}
我将能够访问OpenAI的API密钥,但是用于提取模型列表的代码将返回:{'detail': 'Unauthorized'}
I need the api key enabled from ooba cuz i plan to use the openai.client feature for model interactions. How do I keep the configuration that enables the fake open ai api key but also allows me to extract the model list?
我需要从ooba启用API密钥,因为我计划使用openai.client
功能来进行模型交互。我如何保持这种配置,即启用模拟的OpenAI API密钥,同时又能提取模型列表?
问题解决:
1 day later... Found the answer. Turns out if i want the api key to be enabled while also being able to view the model list from ooba via the api, I need to add an 'Authorization': f'Bearer {api_key}'
in the header.
一天后...我找到了答案。原来,如果我想在启用API密钥的同时,也能够通过API从ooba查看模型列表,我需要在请求头中添加'Authorization': f'Bearer {api_key}'
。
The code should look something like this: 代码应该像这样:
python
import requests
api_key = "enter-your-fake-api-key-here"
url = "http://127.0.0.1:5000/v1"
#Model List
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {api_key}"
}
response = requests.get(f'{url}/internal/model/list',
headers=headers,
verify=False)
print(response.json())