如果你想要在本地构建一个 RAG 系统，我们可以使用 ollama 来做基本的模型，使用 llamaindex 构建 agent。
因为 llamaindex 默认使用 openai，我们需要首先调整默认的 embedding 模型和 llm 模型。
1
2
| Settings.embed_model = OllamaEmbedding(model_name=model_name, base_url=sdmicl[1])
Settings.llm = Ollama(model=sdmicl[0], base_url=sdmicl[1], request_timeout=360.0)
|
base_url 需要替换成你自己的 ollama 实例，例如 http://localhost:11434。
如果目录下的都是 txt 或者 md 数据，可以直接使用 SimpleDirectoryReader 来读取基本的数据。
1
2
| # Create a RAG tool using LlamaIndex
documents = SimpleDirectoryReader("data").load_data()
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
| from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, Settings
from llama_index.embeddings.ollama import OllamaEmbedding
def get_agent(model_name: str):
Settings.embed_model = OllamaEmbedding(model_name=model_name, base_url=sdmicl[1])
Settings.llm = Ollama(model=sdmicl[0], base_url=sdmicl[1], request_timeout=360.0)
# Create a RAG tool using LlamaIndex
documents = SimpleDirectoryReader("data").load_data()
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()
async def search_documents(query: str) -> str:
"""Useful for answering natural language questions about an personal essay written by Paul Graham."""
response = await query_engine.query(query)
return str(response)
agent = FunctionAgent(
name="Agent",
description="Useful for multiplying two numbers and searching documents",
tools=[multiply, search_documents],
llm=ollama,
system_prompt="You are a helpful assistant that can multiply two numbers and search documents to answer questions",
)
return agent
async def main():
models = ('bge-m3', 'nomic-embed-text',)
for model_name in models:
print(f'model: {model_name}')
agent = get_agent(model_name=model_name)
response = await agent.run("What did the paul graham do in college? Also, what's 7 * 8?")
print(str(response))
print("Done.")
print('-' * 100)
await main()
|