Note a compilation of SFT data that supports improvements of math, code, stem, general reasoning, and tool calling capabilities chat 746,622 code 1,896,395 math 2,044,407 stem 20,662,167 tool_calling 310,051 Total 25,659,642 Models used and n_samples: DeepSeek-R1-0528 24,602,969 Qwen3-235B-A22B 1,056,673 Total 25,659,642

Note extension of v1 for SFT and RL data into five target languages: Spanish, French, German, Italian and Japanese. math 239467 code 175000 stem 355000 chat 627720 multilingual_ja 975202 multilingual_de 1015314 multilingual_it 1016503 multilingual_es 935704 multilingual_fr 1001504 Models used and n_samples: DeepSeek-R1-0528 5,713,694 Qwen2.5-14B-Instruct 3,928,913 Qwen3-30B-A3B 627,720 Qwen2.5-32B-Instruct-AWQ 1,015,314 Qwen3-235B-A22B 627,720