Datasets:
UMAP1
float32 -6.43
9.57
| UMAP2
float32 -10.45
10
| token
stringlengths 1
22
| __index_level_0__
int64 0
34.4k
|
|---|---|---|---|
-0.322715 | -1.689318 | the | 0 |
-0.173329 | -1.868714 | , | 1 |
-0.171967 | -1.844381 | . | 2 |
-0.273924 | -1.727005 | of | 3 |
0.044161 | -2.239126 | to | 4 |
-0.176759 | -1.852893 | and | 5 |
-0.314823 | -1.677969 | in | 6 |
-0.324033 | -1.892974 | a | 7 |
-0.512622 | -1.02173 | " | 8 |
-0.298692 | -1.98009 | 's | 9 |
-0.205921 | -1.948954 | for | 10 |
-0.066247 | -2.257194 | - | 11 |
0.151368 | -2.564671 | that | 12 |
-0.223894 | -1.789867 | on | 13 |
-0.144351 | -1.882102 | is | 14 |
-0.463916 | -1.995037 | was | 15 |
-0.573735 | -2.823276 | said | 16 |
-0.235422 | -1.875203 | with | 17 |
-0.665942 | -2.109231 | he | 18 |
-0.275732 | -1.806664 | as | 19 |
0.012048 | -2.081238 | it | 20 |
-0.339077 | -1.892595 | by | 21 |
-0.279424 | -1.65425 | at | 22 |
-1.736592 | 1.450222 | ( | 23 |
-1.661334 | 1.480105 | ) | 24 |
-0.258756 | -1.741043 | from | 25 |
-0.675792 | -2.150754 | his | 26 |
0.63847 | -3.188749 | '' | 27 |
0.732546 | -3.168979 | `` | 28 |
-0.251839 | -1.922748 | an | 29 |
0.148486 | -2.503483 | be | 30 |
-0.273119 | -1.997281 | has | 31 |
-0.202603 | -1.908916 | are | 32 |
-0.294708 | -2.121562 | have | 33 |
0.121907 | -2.257735 | but | 34 |
-0.375965 | -2.015277 | were | 35 |
0.208616 | -2.675909 | not | 36 |
-0.039268 | -2.017441 | this | 37 |
-1.584549 | -2.310716 | who | 38 |
-0.129471 | -2.2982 | they | 39 |
-0.371996 | -2.050686 | had | 40 |
0.749392 | -3.24484 | i | 41 |
-0.217922 | -1.729782 | which | 42 |
0.026093 | -2.625074 | will | 43 |
-0.081197 | -2.272364 | their | 44 |
-1.578534 | 1.443444 | : | 45 |
0.262507 | -1.875333 | or | 46 |
-0.281654 | -1.734304 | its | 47 |
-0.24814 | -1.863167 | one | 48 |
-0.469428 | -1.940689 | after | 49 |
-0.277459 | -1.785448 | new | 50 |
-0.299478 | -2.043839 | been | 51 |
-0.207299 | -1.838227 | also | 52 |
0.664298 | -3.117685 | we | 53 |
0.053392 | -2.681832 | would | 54 |
-2.789424 | -3.692198 | two | 55 |
-0.08047 | -2.061298 | more | 56 |
0.194345 | -2.536235 | ' | 57 |
-0.457309 | -1.842792 | first | 58 |
-0.193733 | -2.076664 | about | 59 |
0.011577 | -1.860304 | up | 60 |
-0.354017 | -2.027341 | when | 61 |
-0.663822 | -2.027336 | year | 62 |
-0.140454 | -1.995239 | there | 63 |
-0.200373 | -1.956988 | all | 64 |
0.041434 | -2.416495 | -- | 65 |
0.117604 | -1.942601 | out | 66 |
-1.737598 | -2.20809 | she | 67 |
-0.173722 | -1.841823 | other | 68 |
-0.402709 | -2.091272 | people | 69 |
0.584865 | -3.158808 | n't | 70 |
-1.965938 | -2.229696 | her | 71 |
6.445725 | 3.501363 | percent | 72 |
-0.125893 | -2.070807 | than | 73 |
-0.031386 | -2.024948 | over | 74 |
0.253656 | -1.510139 | into | 75 |
-0.594161 | -2.001196 | last | 76 |
-0.23286 | -1.997227 | some | 77 |
-0.720036 | -3.018051 | government | 78 |
-0.229986 | -1.859058 | time | 79 |
6.172459 | 3.791836 | $ | 80 |
0.675418 | -3.174816 | you | 81 |
-0.637129 | -1.936204 | years | 82 |
0.272096 | -2.652323 | if | 83 |
0.331109 | -2.656119 | no | 84 |
-0.769943 | -1.693542 | world | 85 |
0.276665 | -2.717502 | can | 86 |
-2.799896 | -3.692402 | three | 87 |
0.607542 | -3.233138 | do | 88 |
-1.603077 | 1.358385 | ; | 89 |
-1.341917 | -2.737084 | president | 90 |
-0.103105 | -1.982847 | only | 91 |
-0.870524 | -2.514168 | state | 92 |
6.210872 | 3.762984 | million | 93 |
0.24405 | -2.597466 | could | 94 |
-0.25329 | -2.299326 | us | 95 |
0.04304 | -2.056041 | most | 96 |
0.02166 | -2.301656 | _ | 97 |
-0.33942 | -2.707002 | against | 98 |
-0.505872 | -2.370224 | u.s. | 99 |
Dataset Card
This dataset is a UMAP 2D-projection of the glove.6B.50d embeddings from Stanford. It is intended as a fast reference for visualizing embeddings in a workshop from the AI Service Center Berlin-Brandenburg at the Hasso Plattner Institute.
Dataset Details
Dataset Description
The embeddings have a vocabulary of 400k tokens with 2 dimensions each token.
Curated by: Mario Tormo Romero
License: cc0-1.0
Dataset Sources
This Dataset has been created with UMAP from the glove.6B.50d embeddings.
Uses
This is a dataset created for pegagogical purposes, and is used in the Working with embeddings Workshop created and organized by the AI Service Center Berlin-Brandenburg at the Hasso Plattner Institute.
Dataset Creation
Curation Rationale
We want to provide with this dataset a fast way of obtaining the required data for our workshops, without having to process the data for long periods during the workshop.
- Downloads last month
- 0