xenova/sponsorblock-ml
Automatically detect in-video YouTube sponsorships, self/unpaid promotions, and interaction reminders.
Uses a two-stage T5 transformer architecture—first extracting relevant transcript segments, then applying a classifier to assign category probabilities—trained on the crowdsourced SponsorBlock database. Deployed as a Streamlit web app with batch prediction capabilities for individual videos, channels, or bulk processing, while also providing evaluation and moderation tools for database quality assessment.
159 stars. No commits in the last 6 months.
Stars
159
Forks
15
Language
Python
License
GPL-3.0
Category
Last pushed
Nov 27, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/xenova/sponsorblock-ml"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
ALucek/NeedleInAVidStack
Extract, timestamp, and analyze specific content from video collections using LLM-powered...
samestrin/llm-newsletter-generator
llm-newsletter-generator transforms a valid RSS feed into a "Newsletter" using AI models via...
Niez-Gharbi/Youtube-Summariser
Summarize your youtube videos with BART on streamlit app.
lyle-mlengineer/timesnap
A web service for extracting timestamps from youtube videos.
lijoraju/llm-news-aggregator
Personalized news summaries using LLMs, FAISS, and Telegram bot.