It is a converted 100k subsample of Tulu-v2-sft-mixture (ODC-BY) using Seq-Instruct method from the paper SIT: Fine-tuning Large Language Models with Sequential Instructions
It includes massive sequential instructions which contains subtask more than one in each instructions.
