coding module:

1. Models: from langchain_ollama/transformers
2. Prompts: from langchain_core.prompts import PromptTemplate
3. Memory: from langchain_community.chat_message_histories import ChatMessageHistory
4. Index:
   1. loader(.loader()): from langchain_unstructured import UnstructuredLoader from langchain_community.document_loaders import TextLoader
   2. splitter(.split_text()): from langchain.text_splitter import CharaterTextSplitter
   3. VectorStore: (l2/ip/cosine) from langchain_community.vectorstores import Chroma
   4. retriever: 需要传入参数query, k, 调用封装好的方法; 如 FAISS 的 as_retriever + invoke; 如 Chroma 的 similarity_search_with_score;

tips:

1. 请手动安装ollama以部署qwen2.5:7B模型，或从huggingface下载模型用transformers库导入
2. 初步安装langchain依赖： pip install langchain langchain_community
3. langchain_community里也有不少好东西，一眼丁真，鉴定为纯纯的小众宝藏