openeurollm/EU-Instruct-Synthetic

EU Instruct Synthetic

Synthetically generated instruction-following SFT data for 11 European languages. Each example is a single-turn chat (messages: a user instruction and an assistant response) with a language field.

This is the synthetic counterpart to openeurollm/Dolci-Instruct-SFT-translated.

Languages and sizes

Code	Language	Examples
cs	Czech	150,129
de	German	135,786
el	Greek	138,048
es	Spanish	132,736
fr	French	119,497
it	Italian	136,407
nl	Dutch	131,515
pl	Polish	146,247
pt	Portuguese	131,707
ro	Romanian	129,112
uk	Ukrainian	146,092
Total		1,497,276

Usage

from datasets import load_dataset

# a single language
es = load_dataset("openeurollm/EU-Instruct-Synthetic", "es")

# all languages combined (default config)
ds = load_dataset("openeurollm/EU-Instruct-Synthetic")

Format

Each row:

{"messages": [{"role": "user", "content": "..."},
              {"role": "assistant", "content": "..."}],
 "language": "es"}

Source

Generated with the synthgen-if synthetic instruction-following pipeline.