Skip to content

Dataset · yandex

5B Scale Music Recommendation Dataset

yandex/yambda

4.79 billion interactions and audio embeddings for music ranking and retrieval.

What's inside

Yambda-5B is an industrial dataset for music recommendations: 4.79 billion interactions of 1 million users with 9.39 million tracks. Includes organic and recommendation interactions as well as audio embeddings. Data is presented in tabular and text form; a benchmark for ranking and retrieval is claimed.

What you can do with it

  • Training and evaluating recommendation ranking models
  • Testing retrieval using audio embeddings
  • Comparing against the Yambda-5B benchmark
Size
1B–10B rows
Downloads this month
43K
Likes
267

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("yandex/yambda")