Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
input_ids
list
[ 27133, 15442, 15442, 20529, 2296, 24788, 15442, 15442, 20529, 2296, 41703, 899, 1456, 6725, 18297, 15442, 1116, 198, 39144, 3191, 198, 198, 3087, 42, 408, 58, 68, 22586, 24, 15051, 25, 198, 198, 35, 34, 33, 5158, 28, 1712, 284, 278, 3...
[ 2355, 338, 457, 719, 3895, 5711, 282, 929, 28, 17392, 359, 37152, 23524, 281, 653, 6389, 2397, 30, 2590, 7563, 281, 18280, 28, 2905, 28, 10026, 28, 14739, 28, 8552, 28, 10976, 7821, 28, 284, 9504, 28, 502, 14376, 3366, 3463, 1062, 3...
[ 288, 8586, 260, 2642, 1564, 3187, 282, 260, 20639, 28, 284, 732, 384, 26615, 314, 253, 12251, 282, 9991, 8660, 28, 5150, 288, 26225, 284, 6857, 13959, 28, 41658, 284, 9403, 9396, 28, 9956, 37260, 284, 8628, 30, 533, 260, 4749, 7305, ...
[ 1350, 357, 430, 496, 375, 28, 2289, 288, 22586, 28, 975, 357, 314, 19971, 38361, 515, 28, 253, 4110, 282, 1062, 5922, 284, 1859, 7946, 535, 28, 6207, 1149, 284, 987, 4612, 4997, 816, 260, 2348, 391, 3161, 282, 30984, 30, 198, 198, ...
[ 1122, 30, 22586, 553, 43080, 26125, 260, 1955, 282, 17392, 28, 429, 817, 20924, 288, 2092, 6032, 28, 284, 260, 10659, 282, 260, 1863, 314, 281, 260, 12162, 1270, 198, 29, 28833, 32635, 28, 9685, 28, 8526, 1638, 5389, 282, 7210, 3760, ...
[ 42, 334, 18412, 13459, 198, 33, 32, 42, 3692, 10092, 284, 22541, 2228, 542, 198, 33, 33, 42, 8095, 5733, 52, 3335, 198, 33, 34, 42, 378, 42472, 391, 11967, 282, 7242, 383, 198, 33, 35, 42, 378, 9129, 282, 28846, 707, 198, 33, 36...
[ 28, 355, 657, 506, 3349, 20014, 9415, 4910, 198, 34, 36, 42, 8325, 92, 1503, 284, 377, 5773, 198, 34, 37, 42, 26323, 350, 1476, 1076, 9659, 4877, 198, 34, 38, 42, 42472, 1687, 284, 43031, 198, 34, 39, 42, 15116, 11712, 284, 16605,...
[ 42, 24752, 14350, 284, 32872, 198, 35, 39, 42, 4896, 5194, 407, 368, 22541, 14350, 198, 35, 40, 42, 422, 17240, 428, 2989, 42, 24708, 284, 35174, 198, 35, 41, 42, 24690, 29361, 42, 3300, 312, 30, 298, 104, 288, 492, 30, 44917, 198...
[ 14195, 39, 33, 4031, 198, 198, 13602, 13892, 260, 24266, 789, 198, 198, 28378, 7770, 17073, 42, 10257, 34, 33, 30, 41, 41, 1885, 34, 40, 31, 14195, 34, 36, 4031, 198, 198, 504, 18655, 9720, 7870, 288, 22541, 14350, 1116, 198, 64, ...
[ 25011, 2477, 754, 4275, 429, 1076, 8430, 395, 30, 1249, 582, 2291, 28, 335, 216, 33, 33, 32, 373, 18612, 28, 3419, 288, 14343, 38306, 624, 26942, 866, 10882, 30, 198, 198, 22695, 7615, 28, 6039, 4157, 37287, 28, 553, 4993, 357, 351,...
[ 32440, 12897, 284, 260, 2582, 3486, 24078, 47785, 5294, 8094, 30, 1116, 16506, 198, 788, 650, 23841, 28, 384, 9648, 260, 1911, 40106, 288, 253, 2061, 9683, 284, 26778, 277, 601, 30, 20677, 260, 2987, 11087, 295, 382, 29, 7444, 69, 701...
[ 650, 1048, 1369, 28, 979, 12457, 351, 650, 2049, 253, 3086, 282, 551, 8222, 17743, 335, 354, 5353, 29863, 29, 10498, 5348, 15754, 30, 198, 198, 14086, 30, 37287, 2312, 2583, 2774, 288, 6524, 260, 1911, 506, 17743, 284, 36580, 28, 527,...
[ 856, 724, 4087, 1911, 1828, 28, 5196, 908, 9681, 601, 30, 198, 198, 18, 49, 2458, 1911, 553, 1837, 28, 858, 290, 5780, 28, 588, 260, 2139, 2007, 553, 253, 9239, 1874, 1627, 384, 1137, 30, 476, 2683, 4875, 357, 284, 1643, 28, 637, ...
[ 260, 2322, 15399, 2693, 1057, 45013, 15033, 6208, 28, 527, 384, 1137, 4502, 1272, 2478, 288, 4571, 1895, 3813, 30, 198, 198, 6502, 30, 4603, 925, 20133, 1131, 28, 617, 10769, 47668, 418, 10135, 5982, 3056, 284, 314, 354, 5851, 9115, 4...
[ 346, 4558, 638, 357, 856, 325, 14005, 30, 1249, 6039, 4157, 314, 1048, 335, 260, 5871, 5595, 282, 451, 7386, 30, 909, 6633, 288, 357, 732, 6093, 416, 982, 28, 975, 384, 4485, 357, 347, 253, 22230, 1270, 198, 198, 6502, 30, 10038, ...
[ 28, 284, 346, 457, 1960, 33666, 3166, 30, 657, 314, 1636, 288, 16823, 355, 829, 1155, 601, 351, 7516, 3751, 30, 1069, 536, 3535, 288, 15793, 30, 1848, 506, 260, 1557, 837, 650, 6596, 25809, 856, 2887, 967, 1502, 282, 1828, 1270, 198...

Tokenized DCLM-Edu for Flash Entmax Training

Fixed-length causal language-model training sequences derived from HuggingFaceTB/dclm-edu at revision dbad8ad71224482740cd9c9d353591adbf62fe04. Text is tokenized with HuggingFaceTB/SmolLM2-135M at revision 93efa2f097d58c2a74874c7e644dbc9b0cee75a2. An EOS token is appended to each document before concatenation and packing.

Documents are assigned deterministically by the first eight bytes of SHA-256 over the source id: bucket 0 of 100 is validation and the other buckets are training. The dataset contains 999,997,440 training tokens and 8,388,608 validation tokens. Configurations 4k and 8k contain the same token streams packed to their named sequence length; 128 is a small smoke-test prefix.

Downloads last month
115