Skip to content

vidconcept-sum generates structured, factual summaries of scientific/educational concepts from video titles or descriptions using an LLM.

Notifications You must be signed in to change notification settings

chigwell/vidconcept-sum

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 

Repository files navigation

vidconcept-sum

PyPI version License: MIT Downloads LinkedIn

vidconcept-sum is a lightweight Python package that turns a video title or short text description into a structured, factual summary of the scientific or educational concept being presented. It leverages an LLM (by default ChatLLM7) to interpret the topic and returns a concise explanation, key terms, and contextual information, all formatted according to a strict regex pattern.


Features

  • One‑line API – just pass a string, get a list of structured summary elements.
  • LLM‑agnostic – works out‑of‑the‑box with ChatLLM7, but you can plug any LangChain‑compatible chat model.
  • Safety‑first – the prompt enforces factual, non‑speculative answers and validates output against a regular expression.
  • Easy installation – available on PyPI.

Installation

pip install vidconcept_sum

Quick Start

from vidconcept_sum import vidconcept_sum

# Simple call using the default ChatLLM7 model (requires an API key)
summary = vidconcept_sum(
    user_input="A short intro to quantum entanglement in YouTube video titles."
)

print(summary)

Parameters

Name Type Description
user_input str The video title or short description to be summarised.
llm (optional) BaseChatModel Any LangChain chat model instance. If omitted, the default ChatLLM7 will be instantiated.
api_key (optional) str API key for LLM7. If not supplied, the function looks for the environment variable LLM7_API_KEY.

The function returns a List[str] containing the extracted summary data.


Using a Custom LLM

You can replace the default ChatLLM7 with any LangChain‑compatible chat model. Below are examples for OpenAI, Anthropic, and Google Generative AI.

OpenAI

from langchain_openai import ChatOpenAI
from vidconcept_sum import vidconcept_sum

llm = ChatOpenAI(model="gpt-4o-mini")
response = vidconcept_sum(
    user_input="Explaining the basics of photosynthesis.",
    llm=llm
)
print(response)

Anthropic

from langchain_anthropic import ChatAnthropic
from vidconcept_sum import vidconcept_sum

llm = ChatAnthropic(model="claude-3-haiku-20240307")
response = vidconcept_sum(
    user_input="A video about black holes and event horizons.",
    llm=llm
)
print(response)

Google Generative AI

from langchain_google_genai import ChatGoogleGenerativeAI
from vidconcept_sum import vidconcept_sum

llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash")
response = vidconcept_sum(
    user_input="Understanding the theory of relativity in simple terms.",
    llm=llm
)
print(response)

API Key & Rate Limits

  • LLM7 free tier provides generous rate limits that cover typical usage of this package.
  • If you need higher limits, supply your own API key:
    • Via environment variable: export LLM7_API_KEY="your_key_here"
    • Or directly in the call: vidconcept_sum(..., api_key="your_key_here")
  • Obtain a free API key by registering at https://token.llm7.io/.

Contributing & Support

If you encounter any issues or have feature requests, please open an issue on GitHub:

https://github.com/chigwell/vidconcept-sum/issues


License

This project is licensed under the MIT License.


Author

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


Enjoy using vidconcept-sum to instantly extract clear, educational concepts from video titles!