Dataset · jamescalam
Technical YouTube Tutorial Transcriptions
jamescalam/youtube-transcriptions
Short text snippets from English video tutorials with timestamps for search and RAG.
What's inside
Contains text transcriptions of technical educational videos from James Briggs, Daniel Bourke, and AI Coffee Break YouTube channels in English. Records were obtained using the OpenAI Whisper (large) model. Each line contains a short text snippet about one sentence long, a video link, and a timestamp. The dataset totals between 100 thousand and 1 million lines.
What you can do with it
- Training and testing semantic search systems and RAG on video content
- Creating question-answering systems linked to video timestamps
- Tasks
- Question answering, Text retrieval, Visual question answering
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 6.5K
- Likes
- 44
How to load it
from datasets import load_dataset
ds = load_dataset("jamescalam/youtube-transcriptions")