Skip to content

Dataset · nyuuzyou

Large corpus of Wildberries customer reviews

nyuuzyou/wb-feedbacks

Between 100 million and 1 billion Russian customer reviews from the Wildberries marketplace in zstd archives with jsonl.

What's inside

A massive Russian-language text corpus containing between 100 million and 1 billion records, gathered by iterating through approximately 230 million product IDs (nmId) on the Wildberries marketplace and extracting available comments. Data is packed in zstd-compressed archives containing jsonl format files.

What you can do with it

  • Pre-training and fine-tuning Russian language models on conversational vocabulary
  • Training sentiment, spam, and product quality aspect classifiers in e-commerce
  • Training models to generate review responses or summarize customer opinions
Tasks
Text generation, Text classification
Languages
Russian
Size
100M–1B rows
Downloads this month
302
Likes
9

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nyuuzyou/wb-feedbacks")