QUEST-SFT-Data-Objective: Synthetic Tool-Use Trajectories for AI Agents

A dataset of objective-style supervised fine-tuning trajectories for QUEST, a tool-using assistant format. Created by osunlp as part of the QUEST family for training deep research agents with fully synthetic tasks. The dataset was last updated on 2026-05-27.

Use Cases

• Fine-tune language models for tool-use capabilities based on the synthetic task trajectories.
• Train research agents to follow objective-style instructions based on the structured message format.
• Benchmark the performance of assistant models on synthetic, tool-using tasks.
• Generate synthetic training data pipelines for agentic AI systems based on the described methodology.

Strengths

• Designed for a specific, advanced AI training purpose: tool-using deep research agents.
• Part of a documented family of datasets (QUEST) with associated project page, paper, and GitHub repository.
• Data structure is explicitly defined with a 'messages' column containing lists of role-content pairs.

Limitations

• Description metadata is limited; actual data quality, size, and specific task scope require manual inspection after download.
• Row count, file formats, and license are unknown, which may limit suitability assessment.
• Data is fully synthetic, which may introduce biases or patterns not present in real-world interactions.

Provenance

Source

osunlp

Collection Method

Synthetically generated for training deep research agents.

Freshness

Last updated 2026-05-27 04:31:05.