dataset stringclasses 2
values | split stringclasses 3
values | file_name stringlengths 5 8 | index int64 1 70k | transcript stringlengths 3 203 | duration_s float64 0.47 44.9 |
|---|---|---|---|---|---|
mnist | train | 0.png | 4 | Nine. | 0.58 |
mnist | train | 1.png | 68 | Zero. | 0.64 |
mnist | train | 3.png | 111 | Three. | 0.795 |
mnist | train | 1501.png | 148 | Mmm, seven. | 1.379 |
mnist | train | 1777.png | 202 | Eight. | 0.794 |
mnist | train | 1779.png | 262 | Two. | 0.767 |
mnist | train | 4.png | 279 | Three. | 0.522 |
mnist | train | 6.png | 303 | Zero. | 1.071 |
mnist | train | 8.png | 384 | Seven. | 0.567 |
mnist | train | 9.png | 411 | Seven. | 1.011 |
mnist | train | 10.png | 422 | Umm, seven. | 2.265 |
mnist | train | 1781.png | 470 | One. | 1.002 |
mnist | train | 11.png | 483 | Seven. | 1.096 |
mnist | train | 12.png | 499 | Eight. | 0.715 |
mnist | train | 1504.png | 514 | Five. | 1.283 |
mnist | train | 1782.png | 544 | Five. | 0.878 |
mnist | train | 1505.png | 547 | Six or four? But I think six. | 4.116 |
mnist | train | 14.png | 561 | Looks like a three. | 2.095 |
mnist | train | 1783.png | 598 | Hmm, six. | 1.846 |
mnist | train | 1506.png | 670 | Three. | 0.862 |
mnist | train | 1507.png | 720 | Umm, eight, maybe? | 2.538 |
mnist | train | 1784.png | 742 | Four. | 0.812 |
mnist | train | 16.png | 768 | Five. | 0.927 |
mnist | train | 1785.png | 778 | I think that's a five. | 3.709 |
mnist | train | 17.png | 798 | Seven. | 0.747 |
mnist | train | 1509.png | 854 | Could be a two, or a three that was cut off at the bottom, but assuming it was not cut off, it's a two. | 9.881 |
mnist | train | 19.png | 925 | Eight. | 0.666 |
mnist | train | 1787.png | 966 | Three. | 0.731 |
mnist | train | 20.png | 994 | Seven. | 1.008 |
mnist | train | 1788.png | 1,030 | Four. | 0.911 |
mnist | train | 21.png | 1,059 | Eight. | 0.867 |
mnist | train | 22.png | 1,062 | Five. | 0.706 |
mnist | train | 1790.png | 1,075 | Oof, erm, looks cut off. Probably a one, potentially seven, but more likely a one. | 8.1 |
mnist | train | 1511.png | 1,088 | Seven. | 0.939 |
mnist | train | 1791.png | 1,127 | Umm, maybe one? Could also be a seven, but, maybe more one. | 9.26 |
mnist | train | 24.png | 1,197 | One. | 0.845 |
mnist | train | 1512.png | 1,201 | Three. | 0.735 |
mnist | train | 1513.png | 1,219 | Uhh, eight. | 2.811 |
mnist | train | 1514.png | 1,239 | Uhh, what is this? An eight, I guess? I mean, hmm, eight? | 14.447 |
mnist | train | 1515.png | 1,244 | Three. | 1.299 |
mnist | train | 27.png | 1,397 | Two. | 1.154 |
mnist | train | 28.png | 1,409 | Three. | 0.682 |
mnist | train | 30.png | 1,422 | Seven. | 0.846 |
mnist | train | 32.png | 1,515 | Six. | 2.139 |
mnist | train | 37.png | 1,641 | Hmm, eight. | 0.924 |
mnist | train | 1793.png | 1,674 | That's a nine. | 1.804 |
mnist | train | 38.png | 1,693 | Four. | 0.858 |
mnist | train | 39.png | 1,699 | One. | 0.698 |
mnist | train | 40.png | 1,762 | Four. | 0.613 |
mnist | train | 41.png | 1,868 | Zero. | 0.731 |
mnist | train | 43.png | 1,929 | Five. | 0.827 |
mnist | train | 1795.png | 1,930 | Umm, either six or zero. Looks a bit more like zero. | 7.424 |
mnist | train | 44.png | 1,965 | Five. | 0.69 |
mnist | train | 1516.png | 1,968 | Nine. | 1.134 |
mnist | train | 1797.png | 2,092 | Three. | 0.717 |
mnist | train | 46.png | 2,175 | Eight. | 0.807 |
mnist | train | 48.png | 2,213 | Umm, seven? | 1.788 |
mnist | train | 1798.png | 2,231 | Eight, but cut off at the bottom. | 2.5 |
mnist | train | 1799.png | 2,286 | Three. | 0.644 |
mnist | train | 50.png | 2,299 | Nine. | 0.734 |
mnist | train | 51.png | 2,300 | Six. | 1.02 |
mnist | train | 52.png | 2,305 | Two. | 1.009 |
mnist | train | 53.png | 2,319 | Seven. | 0.794 |
mnist | train | 54.png | 2,343 | Five. | 0.847 |
mnist | train | 55.png | 2,346 | Two. | 1.03 |
mnist | train | 56.png | 2,446 | One. | 1.525 |
mnist | train | 57.png | 2,473 | Zero. | 0.872 |
mnist | train | 1517.png | 2,554 | Umm, could be a three, could be a five. Umm... Either, really. Umm, I'll go with maybe a five. | 13.496 |
mnist | train | 58.png | 2,583 | One. | 0.941 |
mnist | train | 60.png | 2,686 | Four. | 0.892 |
mnist | train | 61.png | 2,822 | Six. | 0.632 |
mnist | train | 62.png | 2,824 | Two. | 0.601 |
mnist | train | 63.png | 2,860 | Nine. | 1.035 |
mnist | train | 1801.png | 2,862 | Six. | 1.008 |
mnist | train | 65.png | 2,972 | Eight. | 0.712 |
mnist | train | 66.png | 2,981 | Three. | 0.661 |
mnist | train | 1804.png | 3,094 | Eight. | 0.585 |
mnist | train | 68.png | 3,125 | Four. | 1.007 |
mnist | train | 69.png | 3,166 | Two. | 0.959 |
mnist | train | 70.png | 3,206 | One? | 2.383 |
mnist | train | 1805.png | 3,268 | It looks like a six. | 1.854 |
mnist | train | 72.png | 3,273 | Eight, maybe? | 1.416 |
mnist | train | 73.png | 3,302 | Nine. | 0.613 |
mnist | train | 74.png | 3,337 | Zero. | 0.943 |
mnist | train | 1806.png | 3,367 | Umm, six, maybe? Or zero? I'll go with zero, maybe more? | 12.613 |
mnist | train | 76.png | 3,368 | Hmm, that's a seven. | 3.092 |
mnist | train | 77.png | 3,381 | Two. | 1.233 |
mnist | train | 78.png | 3,538 | Seven. | 0.921 |
mnist | train | 79.png | 3,547 | Four. | 0.734 |
mnist | train | 80.png | 3,566 | Nine. | 0.752 |
mnist | train | 82.png | 3,601 | Eight. | 0.729 |
mnist | train | 84.png | 3,646 | Six. | 1.039 |
mnist | train | 85.png | 3,673 | Zero. | 0.745 |
mnist | train | 1807.png | 3,682 | Five. | 0.894 |
mnist | train | 1519.png | 3,692 | One, I think? Maybe seven | 4.345 |
mnist | train | 87.png | 3,737 | Eight. | 0.745 |
mnist | train | 88.png | 3,761 | Four... | 0.987 |
mnist | train | 90.png | 3,841 | Six. | 0.797 |
mnist | train | 1520.png | 3,907 | I think five, though, it can also look like a six. But I think it's a five. | 11.271 |
mnist | train | 91.png | 3,919 | Six. | 0.73 |
End of preview. Expand in Data Studio
Multi-modal extension to the UNCERTAIN-MNIST dataset ( 'image-level-soft-label-digits' repository )
The 'image-level-soft-label-digits' dataset was expanded using a voice-to-text process: viewing the images and classifying them aloud, which was then processed as follows:
- Unfiltered thought process (transcript as string): instead of forcing a single final label, the speech was transcribed word-for-word, intentionally keeping hesitation and hedging such as “umm” or “maybe”
- Duration (in seconds): the time it took to label the digit
- Downloads last month
- 99