metadata
license: apache-2.0
task_categories:
- text-generation
language:
- en
tags:
- kimi
- kimi k3
- synthetic
- reasoning
- thinking
- thought
- CoT
- SFT
- dense
- conversation
- multi-turn
- moonshot-ai
size_categories:
- n<1K
Dataset description
5 complex reasoning and coding requests (generated by deepseek v4 pro and gemini 3.1 pro) and their responses (generated by kimi k3).
Dataset Details
- Curated by: MoreThought
- Funded by: MoreThought
- Shared by: MoreThought
- License: apache 2.0
Uses
Finetuning LLMs or SLMs of any size on advanced kimi k3 distilled reasoning traces.