data: add JupiMind J1R A100x4 12B-token foundation pretrain mix

docs: rewrite dataset card in English to avoid encoding issues

docs: add dataset card, source registry, and license notice

data(pretrain-24k): sync datasets từ outputs vào pretrain/j1r_24k_7b_mix | 2 file | 8.30 GB | .json=1, .jsonl=1 | 22/04/2026 14:47:23

initial commit