Skip to content

Dataset · lordx64

Agentic data with tool calls from Claude Fable-5

lordx64/agentic-distill-fable-5-sft

4,659 English single-turn examples from the closed Claude Fable-5 preview model for agent SFT training.

What's inside

The dataset contains 4,659 English single-turn text pairs collected from the Claude Fable-5 preview model, access to which was discontinued in June 2026. About 81% of the examples (3,793 rows) end with a tool call block (<tool_use>), and the remaining 19% (866 rows) end with a regular text response. Data is packaged in parquet format with a single text column for convenient SFT training.

What you can do with it

  • SFT fine-tuning of models or LoRA on external tool use patterns
  • Creating compact AI agents via preview model behavior distillation
Tasks
Text generation
Languages
English
Size
1K–10K rows
Downloads this month
1.1K
Likes
59

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("lordx64/agentic-distill-fable-5-sft")