https://sdq.kastel.kit.edu/index.php?title=Exploring_The_Robustness_Of_The_Natural_Language_Inference_Capabilties_Of_T5&feed=atom&action=historyExploring The Robustness Of The Natural Language Inference Capabilties Of T5 - Versionsgeschichte2024-03-28T12:23:25ZVersionsgeschichte dieser Seite in SDQ-InstitutsseminarMediaWiki 1.39.6https://sdq.kastel.kit.edu/mediawiki-institutsseminar/index.php?title=Exploring_The_Robustness_Of_The_Natural_Language_Inference_Capabilties_Of_T5&diff=1722&oldid=prevNd2798 am 30. Juni 2021 um 15:26 Uhr2021-06-30T15:26:26Z<p></p>
<table style="background-color: #fff; color: #202122;" data-mw="interface">
<col class="diff-marker" />
<col class="diff-content" />
<col class="diff-marker" />
<col class="diff-content" />
<tr class="diff-title" lang="de">
<td colspan="2" style="background-color: #fff; color: #202122; text-align: center;">← Nächstältere Version</td>
<td colspan="2" style="background-color: #fff; color: #202122; text-align: center;">Version vom 30. Juni 2021, 16:26 Uhr</td>
</tr><tr><td colspan="2" class="diff-lineno" id="mw-diff-left-l5">Zeile 5:</td>
<td colspan="2" class="diff-lineno">Zeile 5:</td></tr>
<tr><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>|betreuer=Jan Keim</div></td><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>|betreuer=Jan Keim</div></td></tr>
<tr><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>|termin=Institutsseminar/2021-07-09</div></td><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>|termin=Institutsseminar/2021-07-09</div></td></tr>
<tr><td class="diff-marker" data-marker="−"></td><td style="color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #ffe49c; vertical-align: top; white-space: pre-wrap;"><div>|kurzfassung=<del style="font-weight: bold; text-decoration: none;">Kurzfassung</del></div></td><td class="diff-marker" data-marker="+"></td><td style="color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>|kurzfassung=<ins style="font-weight: bold; text-decoration: none;">Large language models like T5 perform excellently on various NLI benchmarks. However, it has been shown that even small changes in the structure of these tasks can significantly reduce accuracy. I build upon this insight and explore how robust the NLI skills of T5 are in three scenarios. First, I show that T5 is robust to some variations in the MNLI pattern, while others degenerate performance significantly. Second, I observe that some other patterns that T5 was trained on can be substituted for the MNLI pattern and still achieve good results. Third, I demonstrate that the MNLI pattern translate well to other NLI datasets, even improving accuracy by 13% in the case of RTE. All things considered, I conclude that the robustness of the NLI skills of T5 really depend on which alterations are applied.</ins></div></td></tr>
<tr><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>}}</div></td><td class="diff-marker"></td><td style="background-color: #f8f9fa; color: #202122; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #eaecf0; vertical-align: top; white-space: pre-wrap;"><div>}}</div></td></tr>
</table>Nd2798https://sdq.kastel.kit.edu/mediawiki-institutsseminar/index.php?title=Exploring_The_Robustness_Of_The_Natural_Language_Inference_Capabilties_Of_T5&diff=1692&oldid=prevNd2798: Die Seite wurde neu angelegt: „{{Vortrag |vortragender=Dennis Grötzinger |email=dennis.groetzinger98@gmail.com |vortragstyp=Bachelorarbeit |betreuer=Jan Keim |termin=Institutsseminar/2021-0…“2021-06-09T13:59:24Z<p>Die Seite wurde neu angelegt: „{{Vortrag |vortragender=Dennis Grötzinger |email=dennis.groetzinger98@gmail.com |vortragstyp=Bachelorarbeit |betreuer=Jan Keim |termin=Institutsseminar/2021-0…“</p>
<p><b>Neue Seite</b></p><div>{{Vortrag<br />
|vortragender=Dennis Grötzinger<br />
|email=dennis.groetzinger98@gmail.com<br />
|vortragstyp=Bachelorarbeit<br />
|betreuer=Jan Keim<br />
|termin=Institutsseminar/2021-07-09<br />
|kurzfassung=Kurzfassung<br />
}}</div>Nd2798