【GPT入门】第66 课 llamaIndex调用远程llm模型与embedding模型的方法

https://docs.llamaindex.org.cn/en/stable/api_reference/llms/openai_like/

1. 调用私有模型的方法

1. OpenAILike

OpenAILike 是对 OpenAI 模型的轻量级封装，使其兼容提供 OpenAI 兼容 API 的第三方工具。

官网：
https://docs.llamaindex.org.cn/en/stable/api_reference/llms/openai_like/

pip install llama-index-llms-openai-like

from llama_index.llms.openai_like import OpenAILike

llm = OpenAILike(
    model="my model",
    api_base="https://hostname.com/v1",
    api_key="fake",
    context_window=128000,
    is_chat_model=True,
    is_function_calling_model=False,
)

response = llm.complete("Hello World!")
print(str(response))

2. OpenAILikeEmbedding

https://docs.llamaindex.org.cn/en/stable/api_reference/embeddings/openai_like/

pip install llama-index-embeddings-openai-like

embedding = OpenAILikeEmbedding(
    model_name="my-model-name",
    api_base="https://:1234/v1",
    api_key="fake",
    embed_batch_size=10,
)

2. 调用公开平台的模型

2.1 调用GLM

参考官网：https://docs.bigmodel.cn/cn/guide/develop/http/introduction，　找到api_base，填入下面

from llama_index.llms.openai_like import OpenAILike
llm = OpenAILike(
    model="glm-4",
    api_base="https://open.bigmodel.cn/api/paas/v4/",
    api_key="改为自己的key",
    context_window=128000,
    is_chat_model=True,
    is_function_calling_model=False,
    max_tokens=1024,
    temperature=0.3,
)

response = llm.complete("我是星星之火，我不开心，开导我!")
print(str(response))

执行结果：