r0b0tlab/

qwen3.8-max-glm5.2-kimi-k3-distillation

A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.

	Teachers




	

Teacher Provider Traces

Qwen3.8-Max-Preview Alibaba Cloud Model Studio 48,283

GLM-5.2 Z.AI Coding Plan 5,307

Kimi Code K3 Moonshot AI (Kimi) 4,347… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-glm5.2-kimi-k3-distillation.