token
stringlengths 1
687
| frequency
int64 1
220M
| document_frequency
int64 1
14.8M
|
|---|---|---|
.
| 220,373,313
| 14,840,661
|
the
| 207,397,407
| 14,567,551
|
,
| 200,859,483
| 14,534,043
|
and
| 118,500,132
| 14,417,441
|
to
| 111,384,770
| 14,255,763
|
of
| 99,998,213
| 13,978,837
|
a
| 91,785,542
| 14,012,756
|
in
| 71,784,231
| 13,633,597
|
-
| 63,895,226
| 11,850,610
|
is
| 46,724,328
| 12,292,265
|
for
| 45,215,695
| 12,800,734
|
that
| 37,388,342
| 10,680,109
|
you
| 35,296,967
| 8,028,307
|
’
| 33,518,373
| 7,105,098
|
with
| 33,203,100
| 11,684,661
|
it
| 32,321,629
| 9,910,578
|
on
| 31,732,312
| 11,456,983
|
i
| 31,537,153
| 5,663,010
|
s
| 31,386,672
| 10,086,613
|
are
| 24,034,360
| 9,361,999
|
as
| 23,708,205
| 9,294,975
|
'
| 23,676,541
| 5,501,754
|
this
| 23,674,294
| 9,839,990
|
:
| 23,408,220
| 8,043,709
|
be
| 21,002,999
| 9,043,245
|
)
| 20,137,119
| 7,582,937
|
(
| 19,744,938
| 7,528,213
|
at
| 19,383,777
| 8,984,219
|
your
| 19,138,886
| 5,691,923
|
we
| 18,436,993
| 6,275,863
|
from
| 18,009,409
| 9,022,687
|
or
| 17,814,568
| 7,737,049
|
have
| 17,670,527
| 8,268,437
|
was
| 17,344,065
| 6,130,689
|
by
| 17,000,741
| 8,521,472
|
will
| 15,038,183
| 6,609,774
|
can
| 14,661,083
| 6,919,526
|
"
| 14,653,094
| 2,807,523
|
an
| 14,423,410
| 7,942,320
|
not
| 13,065,275
| 6,850,314
|
but
| 12,376,711
| 6,604,641
|
all
| 12,077,478
| 7,018,950
|
has
| 11,886,125
| 6,533,663
|
they
| 11,498,598
| 5,422,582
|
our
| 11,026,060
| 4,679,621
|
|
| 10,895,938
| 984,090
|
“
| 10,598,066
| 3,637,406
|
he
| 10,550,122
| 3,342,912
|
!
| 10,546,768
| 4,095,201
|
more
| 10,421,121
| 6,238,922
|
one
| 10,418,028
| 6,233,516
|
/
| 10,394,688
| 3,718,439
|
their
| 10,182,690
| 5,272,003
|
t
| 10,151,785
| 5,106,913
|
”
| 10,127,532
| 3,637,736
|
?
| 9,821,153
| 4,376,202
|
my
| 9,677,996
| 3,384,958
|
if
| 9,628,287
| 5,359,733
|
about
| 9,110,966
| 5,467,485
|
so
| 8,621,471
| 5,104,424
|
his
| 8,557,368
| 3,204,607
|
up
| 8,452,780
| 5,320,681
|
new
| 8,253,015
| 4,622,983
|
when
| 8,081,684
| 5,038,845
|
what
| 8,032,937
| 4,664,586
|
out
| 7,973,843
| 5,136,356
|
there
| 7,919,830
| 4,872,671
|
which
| 7,850,967
| 5,022,907
|
time
| 7,798,123
| 4,940,267
|
also
| 7,650,847
| 5,168,564
|
who
| 7,404,174
| 4,509,121
|
been
| 6,580,228
| 4,352,896
|
like
| 6,515,757
| 4,250,715
|
some
| 6,449,519
| 4,268,634
|
do
| 6,372,095
| 3,979,773
|
1
| 6,217,409
| 3,323,973
|
how
| 6,005,782
| 3,759,015
|
other
| 5,918,630
| 4,285,173
|
just
| 5,914,376
| 4,001,459
|
her
| 5,807,644
| 1,863,862
|
get
| 5,715,755
| 3,766,691
|
–
| 5,634,526
| 2,447,664
|
no
| 5,617,169
| 3,846,373
|
had
| 5,613,554
| 3,186,985
|
were
| 5,560,562
| 3,183,054
|
first
| 5,478,123
| 3,747,739
|
its
| 5,363,603
| 3,311,126
|
into
| 5,290,572
| 3,793,377
|
people
| 5,216,094
| 3,087,253
|
them
| 5,213,788
| 3,387,494
|
said
| 5,211,819
| 2,326,739
|
she
| 5,177,020
| 1,700,088
|
would
| 5,173,792
| 3,240,314
|
these
| 5,139,932
| 3,472,946
|
2
| 5,078,554
| 3,014,265
|
year
| 4,995,159
| 3,005,450
|
most
| 4,901,728
| 3,559,344
|
than
| 4,900,150
| 3,557,854
|
over
| 4,883,346
| 3,573,258
|
make
| 4,866,006
| 3,423,046
|
End of preview. Expand
in Data Studio
Fineweb 10Bt sample vocabulary counts
This is the vocabulary of the 10Bt sample of FineWeb. This vocabulary was obtained by normalizing and pretokenizing the vocabulary using the bert-base-uncased tokenizer. You can use this vocabulary to:
- Obtain probabilities of subparts of your corpus.
- Define useful tokenizer extensions without fitting a new tokenizer.
- Analyzing the semantic content of the corpus
The dataset consists of 1.85 million tokens with their associated frequency and document frequency. The dataset is already sorted by frequency, so taking the N top rows also gets you the N most frequent tokens.
Acknowledgments
Thanks Mixedbread AI for a GPU grant for research into small retrieval models.
- Downloads last month
- 3