Skip to content

Dataset · openbmb

500k agent trajectories for SFT

openbmb/UltraData-SFT-Agent-2609

English-Chinese tool use trajectories for instruction-tuning agent models, approx. 500k examples.

What's inside

UltraData-SFT-Agent-2609 features L3-cleaned data for agent instruction-tuning within UltraData. Around 500k examples with executable agent trajectories and tool use; languages are English and Chinese. Created for MiniCPM5-2B post-training and complements UltraData-SFT-2605.

What you can do with it

  • SFT of agent models with tool use
  • MiniCPM5-2B post-training
  • Evaluation and debugging of agent trajectories
Tasks
Text generation, Question answering
Languages
English, Chinese
Size
100K–1M rows
Downloads this month
12.9K
Likes
324

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("openbmb/UltraData-SFT-Agent-2609")