OpenThoughts-Agent-SFT-1K: 1,000 Task and Agent Trajectory Pairs

OpenThoughts-Agent-SFT-1K is a collection of 1,000 (task, agent-trajectory) pairs used for supervised fine-tuning of AI agents. It was created by the open-thoughts organization as part of a scaling series and was last updated on June 8, 2026. This dataset is a component of the OpenThoughts-Agent project, which aims to curate resources for training agents.

Use Cases

• Supervised fine-tuning of language models for agent tasks based on the described task-trajectory pairs.
• Benchmarking agent performance across different scales using the described 1K-example point in a scaling ladder.
• Studying the relationship between task complexity and agent trajectory based on the curated pairs.

Strengths

• Contains 1,000 specific examples, providing a defined scale for experimentation.
• Part of a documented scaling series (316 / 1K / 3.16K / etc.), allowing for controlled scaling studies.
• Explicitly used to train a named model (OpenThinkerAgent-8B-SFT-1K), indicating a practical application.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known but the total size and file formats are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality and example diversity require manual inspection.

Provenance

Source

open-thoughts organization on Hugging Face.

Collection Method

Curated as part of the OpenThoughts-Agent open-source effort.

Freshness

Last updated 2026-06-08 07:00:58; freshness should be verified.