id string | input_ids sequence | num_tokens int64 |
|---|---|---|
<urn:uuid:0d8a309d-25c5-405d-a08a-c11239f0d717> | [
222,
53,
73,
70,
222,
42,
79,
69,
70,
81,
70,
79,
69,
70,
79,
85,
222,
43,
66,
79,
70,
200,
39,
80,
83,
222,
66,
77,
77,
222,
85,
73,
70,
222,
77,
80,
87,
70,
13,
222,
83,
80,
78,
66,
79,
68,
70,
222,
66,
79,
69,
2... | 3,728 |
<urn:uuid:7400301c-e625-46d5-be90-1020cf8d52f8> | [
222,
53,
80,
83,
79,
66,
69,
80,
70,
84,
222,
66,
83,
70,
222,
85,
73,
70,
222,
78,
80,
84,
85,
222,
74,
79,
85,
70,
79,
84,
70,
222,
84,
85,
80,
83,
78,
84,
222,
80,
79,
222,
85,
73,
70,
222,
81,
77,
66,
79,
70,
8... | 2,570 |
<urn:uuid:f166f15d-9976-40ed-8a49-8bed360001ff> | [
222,
42,
84,
222,
85,
73,
74,
84,
222,
67,
80,
79,
70,
222,
66,
222,
47,
70,
66,
79,
69,
70,
83,
85,
73,
66,
77,
222,
71,
77,
86,
85,
70,
32,
200,
36,
66,
87,
70,
222,
35,
70,
66,
83,
222,
71,
70,
78,
86,
83,
222,
... | 10,329 |
<urn:uuid:ca387b2a-7df2-4bb1-8e2d-9e82dcdc8b5a> | [
222,
37,
74,
68,
85,
74,
80,
79,
66,
83,
90,
222,
66,
79,
69,
222,
85,
83,
66,
79,
84,
77,
66,
85,
80,
83,
222,
71,
80,
83,
222,
73,
66,
79,
69,
73,
70,
77,
69,
200,
47,
70,
88,
222,
27,
222,
84,
70,
79,
84,
66,
72... | 23,081 |
<urn:uuid:6aba2b8d-0f86-4d64-b8af-a03c21e98c63> | [
222,
42,
79,
222,
84,
80,
78,
70,
222,
81,
70,
80,
81,
77,
70,
13,
222,
78,
66,
68,
86,
77,
66,
83,
222,
69,
70,
72,
70,
79,
70,
83,
66,
85,
74,
80,
79,
222,
66,
69,
87,
66,
79,
68,
70,
84,
222,
84,
80,
222,
84,
77... | 1,766 |
<urn:uuid:81426048-4862-4bb7-9a74-922f76b4cc47> | [
222,
53,
73,
70,
83,
70,
222,
66,
83,
70,
222,
78,
66,
79,
90,
222,
88,
66,
90,
84,
222,
85,
80,
222,
70,
71,
71,
70,
68,
85,
74,
87,
70,
77,
90,
222,
85,
70,
66,
68,
73,
222,
66,
222,
69,
80,
72,
15,
200,
47,
80,
... | 3,374 |
<urn:uuid:e36683d3-ded6-4e38-af4d-7bc73d1ec7e3> | [
222,
39,
70,
88,
70,
83,
222,
83,
66,
83,
70,
222,
84,
70,
66,
222,
85,
86,
83,
85,
77,
70,
84,
222,
88,
74,
77,
77,
222,
69,
74,
70,
222,
80,
79,
222,
85,
73,
70,
222,
84,
88,
80,
83,
69,
71,
74,
84,
73,
222,
74,
... | 3,510 |
<urn:uuid:4dcad241-2b6b-4970-9112-c67a47a29a2c> | [
222,
34,
222,
67,
86,
77,
77,
80,
68,
76,
222,
68,
66,
83,
85,
222,
80,
83,
222,
80,
89,
222,
68,
66,
83,
85,
222,
74,
84,
222,
66,
222,
85,
88,
80,
14,
88,
73,
70,
70,
77,
70,
69,
222,
80,
83,
222,
71,
80,
86,
83,
... | 1,953 |
<urn:uuid:6c5322ef-413f-4a69-b643-e15509f4014a> | [222,93,35,74,83,85,73,222,79,66,78,70,93,93,47,80,83,78,66,79,222,49,70,83,68,70,87,70,77,222,51,80(...TRUNCATED) | 20,497 |
<urn:uuid:983e5733-551c-4c84-be58-6e25791db2ab> | [222,53,73,70,222,79,66,78,70,84,222,68,73,74,84,70,77,70,69,222,80,79,85,80,222,68,74,85,90,222,85,(...TRUNCATED) | 928 |
End of preview. Expand
in Data Studio
FineWebEDU 20B
A copy of FineWebEDU-20B used for out tokenizer experiments. The subsets are as follows:
bytelevel: the full dataset tokenized using our bytelevel tokenizerbytelevel-subset_1: a 100k-row subset of the bytelevel subset, used to train bytelevel models.bytelevel-subset_2: a 100k-row subset of the bytelevel subset, used to extract llm predictions.bytelevel-llm-data: a copy ofbytelevel-subset_2with lm predictions, used to train bytespan tokenizersbytelevel-subset_3: a 100k-row subset of the bytelevel subset, used to evaluate trained tokenizers
The remaining subsets are all versions of the dataset tokenized with our trained tokenizers:
BPE_64000BPEWP_64000ByteSpanSurprisalMonotonicFrequency_64000ByteSpanSurprisalMonotonicSeeding_64000ByteSpanSurprisalCombinedFrequency_64000ByteSpanSurprisalCombinedSeeding_64000ByteSpanSurprisalGlobalIncrement_64000
- Downloads last month
- 814