Existing tokens: 199,971,619 1/3 - WikiText-103 (fixed API)... Generating test split: 0%| | 0/4358 [00:00 22,635,189 tokens WikiText-103: 100,000 texts -> 45,823,471 tokens WikiText-103: 150,000 texts -> 69,308,297 tokens WikiText-103: 200,000 texts -> 92,105,968 tokens WikiText-103: 250,000 texts -> 115,047,336 tokens WikiText-103: 300,000 texts -> 138,601,321 tokens WikiText-103: 350,000 texts -> 161,788,432 tokens WikiText-103: 400,000 texts -> 184,963,474 tokens WikiText-103: 450,000 texts -> 207,852,194 tokens WikiText-103: 500,000 texts -> 231,214,640 tokens WikiText-103: 550,000 texts -> 254,905,316 tokens WikiText-103: 600,000 texts -> 278,402,058 tokens WikiText-103: 650,000 texts -> 301,290,874 tokens WikiText-103: 700,000 texts -> 324,414,818 tokens WikiText-103: 750,000 texts -> 347,964,383 tokens WikiText-103: 800,000 texts -> 370,967,929 tokens WikiText-103: 850,000 texts -> 394,148,337 tokens WikiText-103: 900,000 texts -> 417,556,771 tokens WikiText-103: 950,000 texts -> 441,075,122 tokens WikiText-103: 1,000,000 texts -> 464,091,223 tokens WikiText-103: 1,050,000 texts -> 487,133,309 tokens WikiText-103: 1,100,000 texts -> 510,482,618 tokens WikiText-103: 1,150,000 texts -> 533,526,405 tokens OK WikiText-103: 540,519,546 tokens 2/3 - C4 English... C4-EN: 50,000 texts -> 106,885,554 tokens OK C4-EN: 215,577,179 tokens 3/3 - FineWeb-Edu... FineWeb-Edu: 50,000 texts -> 242,392,039 tokens ⚠ Received signal 15, will save checkpoint after current step... OK FineWeb-Edu: 480,132,888 tokens TOTAL TOKENS: 1,436,201,232 Train: 1,364,391,170 tokens (1364.4M) Val: 71,810,062 tokens (71.8M) MORE DATA COMPLETE!