llm-structured-summary

llm-structured-summary is a small utility that takes a short piece of text (e.g., a research‑paper title or abstract snippet) and returns a structured summary of the core idea or problem addressed.
The heavy‑lifting is done by an LLM (by default ChatLLM7 from langchain_llm7). The LLM is guided by system & human prompts to produce output that conforms to a predefined XML‑like pattern, making the result easy to parse and reuse.

---

Table of Contents

• Installation
• Quick Start
• Function API
• Using a Custom LLM
• Environment variables & API keys
• Rate limits
• Contributing & Support
• License

---

Installation

pip install llm_structured_summary

---

Quick Start

from llm_structured_summary import llm_structured_summary

# Simple usage – the function will create a ChatLLM7 instance for you
text = "A novel neural architecture for low‑shot learning in computer vision."
summary = llm_structured_summary(user_input=text)

print(summary)          # -> List of extracted XML‑like tags

---

Function API

def llm_structured_summary(
    user_input: str,
    api_key: Optional[str] = None,
    llm: Optional[BaseChatModel] = None,
) -> List[str]:

Parameter	Type	Description
user_input	str	The text that you want to summarise (title, abstract snippet, etc.).
api_key	Optional[str]	API key for the LLM7 service. If omitted, the function will read the LLM7_API_KEY environment variable or fall back to a placeholder value ("None").
llm	Optional[BaseChatModel]	A pre‑configured LangChain chat model. Supplying a custom LLM lets you replace the default ChatLLM7 with OpenAI, Anthropic, Google, etc.

Returns: List[str] – a list of strings that match the regex pattern defined in llmatch_messages.pattern. Each string is a piece of structured data (e.g., <Problem>...</Problem>).

---

Using a Custom LLM

If you prefer to use another provider, simply pass an instantiated LangChain chat model via the llm argument.

OpenAI (ChatGPT)

from langchain_openai import ChatOpenAI
from llm_structured_summary import llm_structured_summary

my_llm = ChatOpenAI(model="gpt-4o-mini")
response = llm_structured_summary(user_input="Your text here", llm=my_llm)
print(response)

Anthropic (Claude)

from langchain_anthropic import ChatAnthropic
from llm_structured_summary import llm_structured_summary

my_llm = ChatAnthropic(model="claude-3-haiku-20240307")
response = llm_structured_summary(user_input="Your text here", llm=my_llm)

Google (Gemini)

from langchain_google_genai import ChatGoogleGenerativeAI
from llm_structured_summary import llm_structured_summary

my_llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash")
response = llm_structured_summary(user_input="Your text here", llm=my_llm)

All of the above examples work because they return a BaseChatModel compatible object, which llm_structured_summary expects.

---

Environment Variables & API Keys

• LLM7_API_KEY – If you do not pass api_key directly, the function will look for this environment variable.
• You can obtain a free API key by registering at https://token.llm7.io/.

export LLM7_API_KEY="your-llm7-api-key"

---

Rate Limits

The free tier of LLM7 provides generous rate limits that are sufficient for most prototyping and research tasks.
If you need higher throughput, upgrade your LLM7 plan or simply switch to another provider (OpenAI, Anthropic, Google) via the Custom LLM section above.

---

Contributing & Support

• Issues / Feature Requests: https://github.com/chigwell/llm_structured_summary/issues
• Pull Requests are welcome – feel free to open a PR after filing an issue.

---

Author

Eugene Evstafev
📧 Email: hi@euegne.plus
🐙 GitHub: chigwell

---

License

This project is licensed under the MIT License. See the LICENSE file for details.