Search is not available for this dataset
input_ids listlengths 2.05k 2.05k |
|---|
[
10000,
770,
7677,
3300,
8430,
1747,
344,
3063,
6047,
3074,
3357,
308,
73,
337,
3340,
370,
3430,
3003,
10,
9700,
1737,
3247,
537,
7007,
343,
1400,
5773,
5643,
3300,
7207,
4077,
5333,
9073,
7770,
7420,
6457,
401,
10001,
10000,
1010,
2302,... |
[
100,
2110,
221,
10001,
10000,
2320,
1161,
1202,
2601,
4042,
1111,
36,
2311,
2026,
1360,
1201,
1410,
120,
2,
3601,
2136,
2102,
2121,
3022,
2123,
261,
301,
241,
1210,
1062,
3316,
5512,
2010,
120,
2021,
2200,
2202,
2066,
1012,
2153,
2501,
... |
[
222,
242,
203,
1020,
2022,
2200,
400,
10001,
10000,
2332,
230,
2222,
2200,
0,
1020,
240,
2210,
4020,
320,
430,
234,
3320,
4020,
400,
3202,
2302,
300,
4024,
1422,
3422,
4043,
7230,
2242,
0,
2032,
3020,
0,
2224,
2200,
432,
3200,
1404,
... |
[
4122,
2101,
143,
3131,
4322,
3223,
1233,
2031,
313,
3212,
3030,
10001,
10000,
3040,
211,
3021,
3321,
3303,
2232,
2320,
312,
3232,
2020,
1303,
3332,
2000,
312,
3223,
2033,
3313,
2312,
2130,
3010,
2223,
2001,
102,
2232,
3012,
3312,
1234,
... |
[
1606,
21,
2636,
6104,
6161,
1600,
1303,
2616,
6606,
2601,
6661,
3212,
8262,
6106,
3611,
10001,
10000,
1311,
3200,
3113,
4613,
606,
1864,
6160,
3106,
6161,
1613,
3616,
1234,
20,
2113,
6130,
4030,
6122,
6331,
2612,
6301,
1163,
3162,
6138,
... |
[
20,
2222,
222,
402,
1110,
2200,
2122,
205,
50,
2110,
2222,
2022,
2121,
460,
6200,
2122,
3,
2200,
2102,
10001,
10000,
7875,
5257,
3879,
5592,
8565,
3855,
8339,
3936,
870,
7583,
5057,
2979,
7385,
3959,
5922,
2998,
5925,
5955,
6582,
6943,
... |
[
107,
1013,
101,
110,
1344,
3,
111,
744,
113,
1701,
4003,
340,
1301,
301,
141,
1010,
1011,
1002,
117,
4010,
1301,
1021,
1101,
10001,
10000,
431,
1111,
4000,
1114,
1050,
4001,
1043,
1043,
3344,
1010,
4001,
110,
30,
131,
114,
444,
140,
... |
[
2022,
200,
2202,
2002,
22,
2120,
201,
2032,
1020,
1020,
2228,
22,
3222,
2202,
32,
202,
1380,
2200,
3021,
200,
1048,
1020,
2202,
330,
2202,
1102,
2308,
10001,
10000,
2202,
310,
3320,
200,
130,
2122,
111,
1202,
222,
1000,
8022,
220,
302... |
[
103,
3331,
3033,
3060,
1133,
1363,
3313,
1310,
3310,
1313,
3303,
1131,
1333,
333,
1301,
3333,
3051,
343,
113,
1303,
3001,
3000,
3301,
3033,
3033,
1831,
1031,
1110,
1133,
6331,
1310,
10001,
10000,
3003,
3333,
14,
1310,
11,
103,
3331,
303... |
[
2131,
11,
3100,
1301,
200,
3001,
3200,
1313,
31,
3100,
2111,
2203,
301,
3100,
1130,
3003,
3320,
1101,
131,
12,
1300,
131,
3000,
3000,
1100,
110,
13,
11,
11,
132,
3023,
3303,
1332,
1023,
23,
10001,
10000,
3330,
1123,
2,
1330,
111,
12... |
[
122,
10001,
10000,
1128,
1100,
1122,
1102,
1112,
1101,
11,
2211,
2011,
11,
211,
8014,
1122,
1122,
52,
1222,
1100,
5015,
111,
1011,
11,
11,
1511,
811,
2000,
102,
1152,
1100,
1120,
222,
2520,
111,
2011,
11,
5201,
2212,
10001,
10000,
164... |
[
2222,
2211,
2000,
2213,
333,
10001,
10000,
3330,
3331,
3220,
2103,
333,
2031,
3033,
2300,
3331,
3233,
3333,
233,
3322,
2232,
3220,
2033,
3333,
3320,
222,
3220,
3323,
333,
222,
3313,
2330,
2302,
3230,
2333,
2322,
312,
3230,
2321,
3133,
2... |
[
13,
100,
2100,
2200,
2,
210,
141,
2000,
3,
10001,
10000,
240,
1001,
4110,
322,
3110,
1001,
20,
1271,
1007,
1001,
303,
441,
1121,
0,
2130,
40,
2000,
4130,
100,
2020,
2,
2110,
111,
2000,
2000,
103,
1210,
2,
1,
41,
441,
1210,
0,
11... |
[
955,
5055,
5515,
5655,
5555,
5354,
5575,
9152,
5215,
5555,
5715,
5935,
1535,
10001,
10000,
6636,
3331,
1613,
6231,
6611,
1136,
6331,
3613,
3333,
3031,
1120,
1656,
6506,
3331,
5561,
3361,
3633,
3513,
3163,
3103,
1111,
3613,
3311,
6113,
311... |
[
0,
4000,
4400,
2442,
10,
2440,
200,
2023,
404,
2004,
4400,
0,
2040,
4020,
1402,
130,
2420,
10001,
10000,
4020,
203,
2023,
244,
21,
2,
2040,
4003,
3044,
1031,
243,
0,
4040,
4002,
4004,
222,
422,
0,
4234,
4002,
0,
4,
2020,
1424,
344... |
[
22,
3533,
2311,
3301,
3601,
3333,
6311,
1333,
3716,
3333,
6235,
3333,
3311,
3311,
1312,
3323,
3231,
3311,
3321,
3211,
2362,
10001,
10000,
1331,
1332,
1332,
1133,
3116,
1336,
3333,
3330,
1632,
3320,
1323,
3633,
3033,
1121,
3303,
6312,
1336... |
[
1,
312,
2300,
3101,
1110,
3301,
3310,
3,
0,
1011,
3023,
33,
133,
3300,
3320,
313,
3,
2101,
0,
3301,
333,
3001,
1010,
1013,
3,
10001,
10000,
1300,
0,
3100,
1313,
3033,
100,
2300,
1331,
2003,
3110,
302,
1310,
3133,
1233,
3302,
1030,
... |
[
5015,
5115,
3551,
1206,
110,
1505,
310,
1552,
5171,
1112,
5515,
10,
1110,
5201,
5113,
1510,
1510,
212,
3011,
6103,
155,
5575,
1010,
1115,
1552,
115,
110,
561,
207,
10001,
10000,
1503,
1116,
1611,
261,
2021,
1530,
2115,
1010,
115,
550,
... |
[
6822,
2111,
8116,
5210,
8211,
8068,
1193,
1818,
1118,
1181,
1118,
1110,
8861,
3212,
1123,
8811,
1518,
8111,
2111,
1111,
1133,
1133,
1132,
1691,
1708,
2133,
1181,
3111,
1111,
3361,
111,
1111,
1111,
10001,
10000,
200,
4,
2020,
700,
3,
3,
... |
[
10000,
4125,
2077,
2518,
5750,
508,
7719,
1785,
2222,
5927,
207,
2105,
1770,
2111,
2002,
2822,
6502,
57,
9222,
2227,
2522,
8051,
2234,
3712,
2282,
7229,
2008,
520,
7873,
2180,
8722,
7955,
1680,
708,
7,
9805,
758,
10001,
10000,
7007,
535... |
End of preview. Expand in Data Studio
PPT Corpus: nca
Raw stage-1 pre-pretraining corpus from the ppt research framework, exported for information-theoretic analysis (e.g. m-local entropy) independent of this repo's training pipeline.
IMPORTANT: token IDs are NOT standard BPE tokens
Neural Cellular Automata grid-rollout patch IDs (positional-numeral encoding of 2x2 color patches, plus start/end tokens), generated by submodules/nca-pre-pretraining.
These input_ids are not decodable with the Pythia (or any other)
tokenizer. They are integers in [0, 10002) with regimen-specific
meaning (see above) — treat this dataset as a corpus of symbol/event
sequences, not text.
Corpus stats
| Field | Value |
|---|---|
| Vocabulary size | 10002 |
Chunk length (seq_len) |
2048 |
| Number of chunks | 160,000 |
| Total tokens | 327,680,000 |
| Generation seed | 42 |
Each row's input_ids is a fixed-length list of 2048 integers — the
exact chunk granularity fed to the model during stage-1 training (formed
by concatenating generated sequences and splitting into non-overlapping
2048-token blocks).
Generation config
{
"grid_size": 12,
"patch_size": 2,
"num_colors": 10,
"d_state": 10,
"num_rules": 5000,
"buffer_size": 256,
"filter_threshold": 0.5,
"dT": 5,
"rollout_steps": 10,
"jax_seed": 42,
"seq_len": 2048
}
Usage
from datasets import load_dataset
ds = load_dataset("sashaboguraev/ppt-nca-corpus")
example = ds["train"][0]["input_ids"] # list[int], length 2048
- Downloads last month
- 46