alaisgood
commented
7 months ago 找到原因了在from pylmkit.llms init.py里少写了一个括号。 你看有没有必要再增加一个Nvidia的模型。 import os from openai import OpenAI
class ChatNvidia(object): def init(self, nvidia_apikey="", model="meta/llama2-70b", temperature=0.6, top_p=0.7): self.api_key = os.environ.get("nvidia_apikey", nvidia_apikey) self.model = model self.temperature = temperature self.top_p = top_p
self.max_tokens = max_tokens
def invoke(self, query, system_prompt="You need to answer user questions", **kwargs):
client = OpenAI(base_url = "https://integrate.api.nvidia.com/v1", api_key = self.api_key) # 请填写您自己的APIKey
response = client.chat.completions.create(
model=self.model, # 填写需要调用的模型名称
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": query}
],
stream=False,
temperature=self.temperature,
top_p=self.top_p,
max_tokens=1024,
**kwargs
)
return response.choices[0].message.content
def stream(self, query, system_prompt="You need to answer user questions", **kwargs):
client = OpenAI(base_url = "https://integrate.api.nvidia.com/v1", api_key = self.api_key) # 请填写您自己的APIKey
response = client.chat.completions.create(
model=self.model, # 填写需要调用的模型名称
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": query}
],
stream=True,
temperature=self.temperature,
top_p=self.top_p,
max_tokens=1024,
**kwargs
)
for chunk in response:
yield chunk.choices[0].delta.content 
xiaozhi-agent
我参考你的code自己也写了一个LLM class， 但是运行时显示 ChatNvidia.invoke() missing 1 required positional arguments: 'query' \pylmkit\app\roleplay.py", line 31, in invoke response = self.model.invoke() File "C:\Users\dscshap1033\Documents\VSPython\python-3.11.8-embed-amd64\Lib\site-packages\pylmkit\core\base.py", line 322, in run result = obj(**self.input_kwargs)
请问新增LLM,除了pylmkit\llms下增加了class，还涉及什么地方的修改？
class ChatNvidia(object): def init(self, nvidia_apikey="", model="meta/llama2-70b", temperature=0.6, top_p=0.7): self.api_key = os.environ.get("nvidia_apikey", nvidia_apikey) self.model = model self.temperature = temperature self.top_p = top_p
self.max_tokens = max_tokens