Build a Cultural Name‑Generator with LangChain, Custom Prompts, and Output Parsers

Installation

Install the required LangChain version:

pip install --upgrade langchain==0.0.279 -i https://pypi.org/simple

1. Create an LLM

Use your own compute with an open‑source large model (requires substantial GPU resources) and your own training data.

Or call a third‑party LLM API such as OpenAI, Baidu Wenxin, or Alibaba Tongyi; data preparation is optional.

Example uses OpenAI’s gpt-3.5‑turbo‑instruct model.

2. Custom Prompt Template

Parameterize the prompt so the same template can generate names for different cultures.

Support passing variables (e.g., {county}, {boy}, {girl}).

Sample template:

"You are a naming master, generate three {county} names, e.g., boy name {boy}, girl name {girl}."

Usage

Import the template class:

from langchain.prompts import PromptTemplate

3. Output Parser

Convert the LLM’s raw text into a structured format such as a JSON array or a comma‑separated list.

Implement a subclass of BaseOutputParser that splits the result on commas.

from langchain.schema import BaseOutputParser

class CommaSeparatedListOutputParser(BaseOutputParser):
    """Parse the output of an LLM call to a comma‑separated list."""
    def parse(self, text: str):
        return text.strip().split(", ")

# Example
print(CommaSeparatedListOutputParser().parse("hi, bye"))

4. Full Working Example

Set environment variables for the OpenAI key and optional proxy:

import os
os.environ["OPENAI_KEY"] = "xxxxx"
os.environ["OPENAI_API_BASE"] = "xxxxx"  # if a proxy is needed

Load the key in Python:

import os
openai_api_key = os.getenv("OPENAI_KEY")
openai_api_base = os.getenv("OPENAI_API_BASE")
print("OPENAI_API_KEY:", openai_api_key)
print("OPENAI_PROXY:", openai_api_base)

Instantiate the LLM and the prompt:

from langchain.llms import OpenAI
from langchain.prompts import PromptTemplate

llm = OpenAI(
    model="gpt-3.5-turbo-instruct",
    temperature=0,
    openai_api_key=openai_api_key,
    openai_api_base=openai_api_base,
)

prompt = PromptTemplate.from_template(
    "You are a naming master, generate three {county} names. Example: boy {boy}, girl {girl}. Return a comma‑separated list only."
)
message = prompt.format(county="Chinese", boy="DogEgg", girl="CuiHua")
print(message)

raw_output = llm.predict(message)
print(raw_output)

parser = CommaSeparatedListOutputParser()
names = parser.parse(raw_output)
print(names)

The script prints a list such as ['Jack', 'Michael', 'Jason'], demonstrating how to obtain structured name data.

5. Verify Installation

!pip show langchain
!pip show openai

Images in the original article illustrate the workflow and can be referenced for visual guidance.