Acta: A Premium Sample of 5,000 High-Quality Agentic Tool-Use Conversations

A public sample of 5,000 conversations from the full Acta dataset, last updated on 2026-04-21. This dataset was created by DJLougen and contains a curated selection of agentic tool-use dialogues filtered using an 8-factor quality model based on statistical correlation analysis of diversity metrics.

Use Cases

• Training agentic language models based on high-quality, multi-turn tool-use conversations.
• Benchmarking agent performance using a curated sample filtered for conversation quality.
• Analyzing patterns in agentic behavior based on evidence-based quality metrics.
• Studying the diversity of agentic interactions across different source datasets.

Strengths

• Contains 5,000 high-quality conversations from a larger 37,000-conversation dataset.
• Filtered using an 8-factor quality model based on statistical correlation analysis.
• Curated from 5 major source datasets (FlameFox, Hermes Reasoning, Hermes Multi-turn, Smolagents, SWE Agent).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full dataset is known, but the sample's exact composition and column details are unspecified.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Curated from 5 major agentic datasets: FlameFox, Hermes Reasoning, Hermes Multi-turn, Smolagents, SWE Agent.

Collection Method

Filtered using an 8-factor quality model based on statistical correlation analysis of diversity metrics.

Freshness

Last updated 2026-04-21 02:03:39; freshness should be verified.