https://datafranca.org/wiki/index.php?action=history&feed=atom&title=Locally_typical_samplingLocally typical sampling - Historique des versions2026-04-10T18:45:42ZHistorique des versions pour cette page sur le wikiMediaWiki 1.39.5https://datafranca.org/wiki/index.php?title=Locally_typical_sampling&diff=118681&oldid=prevArianne le 5 mars 2026 à 16:292026-03-05T16:29:17Z<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="fr">
<td colspan="2" style="background-color: #fff; color: #202122; text-align: center;">← Version précédente</td>
<td colspan="2" style="background-color: #fff; color: #202122; text-align: center;">Version du 5 mars 2026 à 12:29</td>
</tr><tr><td colspan="2" class="diff-lineno" id="mw-diff-left-l2">Ligne 2 :</td>
<td colspan="2" class="diff-lineno">Ligne 2 :</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;"><br/></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;"><br/></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>== Définition ==</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>== Définition ==</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><del style="font-weight: bold; text-decoration: none;">xxxxxxx</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><ins style="font-weight: bold; text-decoration: none;">Méthode d''''[[échantillonnage]]''' qui améliore la qualité du texte et réduit les répétitions des '''[[Génération de texte|générateurs de texte]]''' '''[[Échantillon aléatoire|probabilistes]]'''. L'idée est que, pour qu'un texte soit perçu comme naturel, chaque mot doit avoir un contenu informatif proche de celui attendu d'après le contexte antérieur. Ainsi, lors de l'échantillonnage à partir de générateurs de texte probabilistes, les options sont limitées aux chaînes de caractères qui respectent cette propriété.</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;"><br/></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;"><br/></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>Voir aussi '''[[échantillonnage des k-meilleurs]]''' et '''[[ergodicité]]'''</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>Voir aussi '''[[échantillonnage des k-meilleurs]]''' et '''[[ergodicité]]'''</div></td></tr>
<tr><td colspan="2" class="diff-lineno" id="mw-diff-left-l12">Ligne 12 :</td>
<td colspan="2" class="diff-lineno">Ligne 12 :</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>''' Locally typical sampling'''</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>''' Locally typical sampling'''</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;"><br/></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;"><br/></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><!--<del style="font-weight: bold; text-decoration: none;">Sampling </del>method that improves text quality and reduces repetitions in probabilistic language generators. The idea is that when you want a text to be perceived as natural, each word should have an information content close to its expected information content given prior context. Thus, when sampling from probabilistic language generators, we should limit our options to strings that adhere to this property.--></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><!--<ins style="font-weight: bold; text-decoration: none;">A sampling </ins>method that improves text quality and reduces repetitions in probabilistic language generators. The idea is that when you want a text to be perceived as natural, each word should have an information content close to its expected information content given prior context. Thus, when sampling from probabilistic language generators, we should limit our options to strings that adhere to this property.--></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;"><br/></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;"><br/></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>== Sources ==</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>== Sources ==</div></td></tr>
</table>Ariannehttps://datafranca.org/wiki/index.php?title=Locally_typical_sampling&diff=118461&oldid=prevArianne : Page créée avec « == en construction == == Définition == xxxxxxx Voir aussi '''échantillonnage des k-meilleurs''' et '''ergodicité''' == Français == ''' XXXXXX''' == Anglais == ''' Locally typical sampling''' <!--Sampling method that improves text quality and reduces repetitions in probabilistic language generators. The idea is that when you want a text to be perceived as natural, each word should have an information content close to its expected information conte... »2026-02-22T20:07:37Z<p>Page créée avec « == en construction == == Définition == xxxxxxx Voir aussi '''<a href="/wiki/%C3%89chantillonnage_des_k-meilleurs" class="mw-redirect" title="Échantillonnage des k-meilleurs">échantillonnage des k-meilleurs</a>''' et '''<a href="/wiki/Ergodicit%C3%A9" title="Ergodicité">ergodicité</a>''' == Français == ''' XXXXXX''' == Anglais == ''' Locally typical sampling''' <!--Sampling method that improves text quality and reduces repetitions in probabilistic language generators. The idea is that when you want a text to be perceived as natural, each word should have an information content close to its expected information conte... »</p>
<p><b>Nouvelle page</b></p><div>== en construction ==<br />
<br />
== Définition ==<br />
xxxxxxx<br />
<br />
Voir aussi '''[[échantillonnage des k-meilleurs]]''' et '''[[ergodicité]]'''<br />
<br />
== Français ==<br />
''' XXXXXX'''<br />
<br />
== Anglais ==<br />
''' Locally typical sampling'''<br />
<br />
<!--Sampling method that improves text quality and reduces repetitions in probabilistic language generators. The idea is that when you want a text to be perceived as natural, each word should have an information content close to its expected information content given prior context. Thus, when sampling from probabilistic language generators, we should limit our options to strings that adhere to this property.--><br />
<br />
== Sources ==<br />
[https://arxiv.org/pdf/2202.00666 Source : arxiv]<br />
<br />
[https://huggingface.co/papers/2202.00666 Source : Hugging Face]<br />
<br />
[[Catégorie:vocabulary]]</div>Arianne