Datasets:

danish-foundation-models
/

danish-wildchat4.8M

Modalities:

Formats:

Languages:

Size:

Tags:

Libraries:

Dataset card Data Studio Files Files and versions

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Danish WildChat 4.8

This dataset contains Danish translations of the first non-empty user message from each conversation in allenai/WildChat-4.8M.

Translation setup

Source dataset: allenai/WildChat-4.8M (train split)
Target dataset repo: danish-foundation-models/danish-wildchat4.8M
Translation model: google/translategemma-27b-it
Translation backend: vLLM
Target language: Danish (da)
Upload pattern: one Parquet shard per configured successful translations

Row schema

Each row stores the original message, the translated Danish message, the resolved source language code, the translation model id, and selected WildChat metadata such as conversation_hash and accept_language.

Downloads last month: 798

Total file size: