<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>https://air.imag.fr/index.php?action=history&amp;feed=atom&amp;title=VT2021_GPT_fiche</id>
	<title>VT2021 GPT fiche - Revision history</title>
	<link rel="self" type="application/atom+xml" href="https://air.imag.fr/index.php?action=history&amp;feed=atom&amp;title=VT2021_GPT_fiche"/>
	<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;action=history"/>
	<updated>2026-06-20T01:09:47Z</updated>
	<subtitle>Revision history for this page on the wiki</subtitle>
	<generator>MediaWiki 1.39.17</generator>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51982&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Résumé */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51982&amp;oldid=prev"/>
		<updated>2022-01-15T12:32:31Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Résumé&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 12:32, 15 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51981&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Résumé */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51981&amp;oldid=prev"/>
		<updated>2022-01-15T12:32:07Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Résumé&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 12:32, 15 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-added&quot;&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51980&amp;oldid=prev</id>
		<title>Oumaima.Hajji at 12:31, 15 January 2022</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51980&amp;oldid=prev"/>
		<updated>2022-01-15T12:31:50Z</updated>

		<summary type="html">&lt;p&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 12:31, 15 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 5:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Mots Clés&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Mots Clés&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;: GPT, Generative Pre-trained Transformer, est un réseau de neurones basé sur l’architecture “Transformers”.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;: GPT, Generative Pre-trained Transformer, est un réseau de neurones basé sur l’architecture “Transformers”.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;TAL&#039;&#039;&#039;: TAL est l’ensemble des techniques informatiques mises en œuvre pour traiter le langage naturel (oral ou écrit) par une machine.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;TAL&#039;&#039;&#039;: TAL est l’ensemble des techniques informatiques mises en œuvre pour traiter le langage naturel (oral ou écrit) par une machine.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: est un modèle d&#039;apprentissage profond utilisé dans le domaine TAL qui facilite la parallélisation lors du traitement des données. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: est un modèle d&#039;apprentissage profond utilisé dans le domaine TAL qui facilite la parallélisation lors du traitement des données. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Abstract&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Abstract&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT is a network of neurons that is based on the &quot;Transformers&quot; architecture. It&#039;s a new approach to natural language processing that aims to succeed old models that are more costly and less efficient. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT is a network of neurons that is based on the &quot;Transformers&quot; architecture. It&#039;s a new approach to natural language processing that aims to succeed old models that are more costly and less efficient. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;The first series GPT (GPT-1, GPT-2, &amp;amp; GPT-3) is the result of an enormous amount of research ad development done by OpenAI. It&#039;s thanks to this series that the doors opened to other contributors to advance the field of NLP, especially EleutherAI and the model GPT-J.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;The first series GPT (GPT-1, GPT-2, &amp;amp; GPT-3) is the result of an enormous amount of research ad development done by OpenAI. It&#039;s thanks to this series that the doors opened to other contributors to advance the field of NLP, especially EleutherAI and the model GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt; &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt; &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Keywords&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Keywords&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;:  GPT, Generative Pre-trained Transformer, is a network of neurons that is based on the &quot;Transformers&quot; architecture and used in the fields of natural language processing.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;:  GPT, Generative Pre-trained Transformer, is a network of neurons that is based on the &quot;Transformers&quot; architecture and used in the fields of natural language processing.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;NLP&#039;&#039;&#039;: NLP, or Natural Language Processing, is an array of techniques used to process and analyze natural language (either written or spoken) by a machine.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;NLP&#039;&#039;&#039;: NLP, or Natural Language Processing, is an array of techniques used to process and analyze natural language (either written or spoken) by a machine.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: It&#039;s a deep learning model that adopts the mechanism of self-attention and differentially weighs the significance of each part of the input data. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: It&#039;s a deep learning model that adopts the mechanism of self-attention and differentially weighs the significance of each part of the input data. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51979&amp;oldid=prev</id>
		<title>Oumaima.Hajji at 12:30, 15 January 2022</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51979&amp;oldid=prev"/>
		<updated>2022-01-15T12:30:49Z</updated>

		<summary type="html">&lt;p&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 12:30, 15 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 2:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 2:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;* AGUIAR Mathilde ([[mathilde.aguiar@etu.univ-grenoble-alpes.fr]])&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;* AGUIAR Mathilde ([[mathilde.aguiar@etu.univ-grenoble-alpes.fr]])&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;* HAJJI Oumaima ([[oumaima.hajji@etu.univ-grenoble-alpes.fr]])&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;* HAJJI Oumaima ([[oumaima.hajji@etu.univ-grenoble-alpes.fr]])&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Résumé&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT est un réseau de neurones basé sur l&#039;architecture &quot;Transformers&quot;. Ce réseau se manifeste comme une nouvelle approche dans le domaine TAL (Traitement automatique des langues) qui succède à d&#039;autres modèles beaucoup plus coûteux et moins performants. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;La première série GPT (GPT-1, GPT-2, et GPT-3) est le fruit de recherches et de développement de la société OpenAI qui a introduit ce modèle dans le domaine TAL et qui a ouvert la porte à d&#039;autres contributeurs, notamment EleutherAI et le modèle GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Mots Clés&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;: GPT, Generative Pre-trained Transformer, est un réseau de neurones basé sur l’architecture “Transformers”.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;TAL&#039;&#039;&#039;: TAL est l’ensemble des techniques informatiques mises en œuvre pour traiter le langage naturel (oral ou écrit) par une machine.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: est un modèle d&#039;apprentissage profond utilisé dans le domaine TAL qui facilite la parallélisation lors du traitement des données. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Abstract&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT is a network of neurons that is based on the &quot;Transformers&quot; architecture. It&#039;s a new approach to natural language processing that aims to succeed old models that are more costly and less efficient. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;The first series GPT (GPT-1, GPT-2, &amp;amp; GPT-3) is the result of an enormous amount of research ad development done by OpenAI. It&#039;s thanks to this series that the doors opened to other contributors to advance the field of NLP, especially EleutherAI and the model GPT-J.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt; &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Keywords&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;GPT&#039;&#039;&#039;:  GPT, Generative Pre-trained Transformer, is a network of neurons that is based on the &quot;Transformers&quot; architecture and used in the fields of natural language processing.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;NLP&#039;&#039;&#039;: NLP, or Natural Language Processing, is an array of techniques used to process and analyze natural language (either written or spoken) by a machine.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;&#039;&#039;&#039;Transformer&#039;&#039;&#039;: It&#039;s a deep learning model that adopts the mechanism of self-attention and differentially weighs the significance of each part of the input data. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Introduction au Traitement automatique des Langues (TAL)&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Introduction au Traitement automatique des Langues (TAL)&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51871&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Le plus récent, GPT-3 */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51871&amp;oldid=prev"/>
		<updated>2022-01-03T11:11:03Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Le plus récent, GPT-3&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 11:11, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3 [&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;10&lt;/del&gt;]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3 [&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;9&lt;/ins&gt;]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle [&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;11&lt;/del&gt;]. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle [&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;10&lt;/ins&gt;]. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51870&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Performance */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51870&amp;oldid=prev"/>
		<updated>2022-01-03T11:10:38Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Performance&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 11:10, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 99:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 99:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-2.PNG|Figure 7: Comparaison des performances de GPT-J et des autres modèles ]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-2.PNG|Figure 7: Comparaison des performances de GPT-J et des autres modèles ]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 7: Comparaison des performances de GPT-J et des autres modèles [&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;9&lt;/del&gt;]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 7: Comparaison des performances de GPT-J et des autres modèles [&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;8&lt;/ins&gt;]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le plus récent, GPT-3&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le plus récent, GPT-3&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51869&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Bibliographie */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51869&amp;oldid=prev"/>
		<updated>2022-01-03T11:10:19Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Bibliographie&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 11:10, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 162:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 162:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Bibliographie&#039;&#039;&#039; =&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Bibliographie&#039;&#039;&#039; =&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[1] Attention Is All You Need: https://arxiv.org/abs/1706.03762&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[2] The true names/sizes of the 4 GPT-2 models (Issue): https://github.com/openai/gpt-2/issues/209&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[3] Language Models are Unsupervised Multitask Learners : https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[4] Le projet sur github pour plus d’informations: https://github.com/kingoflolz/mesh-transformer-jax/&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[5] Documentation de GPT-J sur Huggingface: https://huggingface.co/docs/transformers/model_doc/gptj&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[6] Documentation de la pile: https://pile.eleuther.ai/&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[7] The Pile: An 800GB Dataset of Diverse Text for Language Modeling: &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;https://arxiv.org/pdf/2101.00027.pdf&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[8] GPT-J-6B: 6B JAX-Based Transformer (Aran Komatsuzaki): https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[9] Language Models are Few-Shot Learners: https://arxiv.org/pdf/2005.14165.pdf&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-empty diff-side-deleted&quot;&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[10] How Biased is GPT-3: https://medium.com/fair-bytes/how-biased-is-gpt-3-5b2b91f1177&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51867&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Le plus récent, GPT-3 */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51867&amp;oldid=prev"/>
		<updated>2022-01-03T10:58:42Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Le plus récent, GPT-3&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 10:58, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3 [&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;4&lt;/del&gt;]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3 [&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;10&lt;/ins&gt;]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle [&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;5&lt;/del&gt;]. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle [&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;11&lt;/ins&gt;]. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51866&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Le modèle open-source, GPT-J */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51866&amp;oldid=prev"/>
		<updated>2022-01-03T10:58:17Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Le modèle open-source, GPT-J&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 10:58, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 73:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 73:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le modèle open-source, GPT-J&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le modèle open-source, GPT-J&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Un mot sur GPT-J&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Un mot sur GPT-J&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;C’est un modèle de 6B paramètres utilisant l’approche transformer pour générer du texte. C’est un projet Open Source de Eleuther.ai, publié sur github[&lt;del style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;*&lt;/del&gt;] par Ben Wang and Aran Komatsuzaki. GPT-J a été entraîné avec la dataset pile de Eleuther.ai.  &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;C’est un modèle de 6B paramètres utilisant l’approche transformer pour générer du texte. C’est un projet Open Source de Eleuther.ai, publié sur github[&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;4&lt;/ins&gt;] par Ben Wang and Aran Komatsuzaki. GPT-J a été entraîné avec la dataset pile de Eleuther.ai&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [5]&lt;/ins&gt;.  &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Le dataset Pile&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Le dataset Pile&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Cette dataset (825 GB) respecte les principes d’une bonne dataset dont on a parlé dans la partie GPT-2. En effet, elle contient des données diversifiée qui couvrent plusieurs domaines ce qui la rend la meilleure dataset pour entraîner un modèle général et pour pouvoir exécuter des différentes tâches. &amp;lt;/br&amp;gt;&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Cette dataset (825 GB) respecte les principes d’une bonne dataset dont on a parlé dans la partie GPT-2. En effet, elle contient des données diversifiée qui couvrent plusieurs domaines ce qui la rend la meilleure dataset pour entraîner un modèle général et pour pouvoir exécuter des différentes tâches&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [6]&lt;/ins&gt;. &amp;lt;/br&amp;gt;&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Elle se compose de 22 sous datasets qui rentrent dans des différents domaines. Les documents d’une dataset peuvent être revus plusieurs fois ‘epochs’ durant un ‘epoch’ sur la dataset entière. Par exemple, puisque Wikipedia est considérée comme une source des documents de bonne qualité, son ‘epoch’ est vu 3 fois plus que d’autres datasets.&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Elle se compose de 22 sous datasets qui rentrent dans des différents domaines. Les documents d’une dataset peuvent être revus plusieurs fois ‘epochs’ durant un ‘epoch’ sur la dataset entière. Par exemple, puisque Wikipedia est considérée comme une source des documents de bonne qualité, son ‘epoch’ est vu 3 fois plus que d’autres datasets.&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-1.PNG|Figure 5: Les sous datasets de la Pile]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-1.PNG|Figure 5: Les sous datasets de la Pile]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 5: Les sous datasets de la Pile&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 5: Les sous datasets de la Pile&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [7]&lt;/ins&gt;&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Utilisation du modèle&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Utilisation du modèle&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT-J a été particulièrement développé pour améliorer la génération des textes suivant les instructions données. En fournissant une ‘prompt’ en entrée, GPT-J renvoie un texte en respectant les paramètres donnés. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;GPT-J a été particulièrement développé pour améliorer la génération des textes suivant les instructions données. En fournissant une ‘prompt’ en entrée, GPT-J renvoie un texte en respectant les paramètres donnés. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Il répond bien à des prompts de différents types, par exemple :&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Il répond bien à des prompts de différents types, par exemple &lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt;[8]&lt;/ins&gt;:&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-3.PNG|Figure 6: exemple de &#039;prompt&#039; donnée en entrée à GPT-J]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-3.PNG|Figure 6: exemple de &#039;prompt&#039; donnée en entrée à GPT-J]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 99:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 99:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-2.PNG|Figure 7: Comparaison des performances de GPT-J et des autres modèles ]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:Gptj-2.PNG|Figure 7: Comparaison des performances de GPT-J et des autres modèles ]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 7: Comparaison des performances de GPT-J et des autres modèles&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 7: Comparaison des performances de GPT-J et des autres modèles&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [9]&lt;/ins&gt;&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le plus récent, GPT-3&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;= &#039;&#039;&#039;Le plus récent, GPT-3&#039;&#039;&#039; = &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
	<entry>
		<id>https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51865&amp;oldid=prev</id>
		<title>Oumaima.Hajji: /* Le plus récent, GPT-3 */</title>
		<link rel="alternate" type="text/html" href="https://air.imag.fr/index.php?title=VT2021_GPT_fiche&amp;diff=51865&amp;oldid=prev"/>
		<updated>2022-01-03T10:56:26Z</updated>

		<summary type="html">&lt;p&gt;&lt;span dir=&quot;auto&quot;&gt;&lt;span class=&quot;autocomment&quot;&gt;Le plus récent, GPT-3&lt;/span&gt;&lt;/span&gt;&lt;/p&gt;
&lt;table style=&quot;background-color: #fff; color: #202122;&quot; data-mw=&quot;interface&quot;&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;col class=&quot;diff-marker&quot; /&gt;
				&lt;col class=&quot;diff-content&quot; /&gt;
				&lt;tr class=&quot;diff-title&quot; lang=&quot;en&quot;&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;← Older revision&lt;/td&gt;
				&lt;td colspan=&quot;2&quot; style=&quot;background-color: #fff; color: #202122; text-align: center;&quot;&gt;Revision as of 10:56, 3 January 2022&lt;/td&gt;
				&lt;/tr&gt;&lt;tr&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
  &lt;td colspan=&quot;2&quot; class=&quot;diff-lineno&quot;&gt;Line 132:&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;[[File:ConfigGPT3.png| Figure 9: informations sur les différents modèles de GPT-3]]&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Figure 9: informations sur les différents modèles de GPT-3&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [4]&lt;/ins&gt;&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Limitations&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Malgré le fait que GPT-3 atteigne d&#039;excellents scores sur de nombreuses tâches, le modèle possède certaines limites. Tout d’abord, il est plutôt faible sur les tâches de synthèse de textes, il fait beaucoup de répétitions et même des contradictions. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;−&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle. &lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot; data-marker=&quot;+&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;Un autre problème récurrent chez les modèles de deep learning est le biais, et GPT-3 n’y échappe pas non plus. Certains utilisateurs ont pu constater des biais racistes ou misogynes comme par exemple l’association du mot “Islam” avec “terrorisme” qui revient souvent dans les phrases générées par le modèle&lt;ins style=&quot;font-weight: bold; text-decoration: none;&quot;&gt; [5]&lt;/ins&gt;. &lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;br /&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
  &lt;td class=&quot;diff-marker&quot;&gt;&lt;/td&gt;
  &lt;td style=&quot;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;&quot;&gt;&lt;div&gt;==&#039;&#039;&#039;Autres modèles issus de GPT-3&#039;&#039;&#039;==&lt;/div&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/table&gt;</summary>
		<author><name>Oumaima.Hajji</name></author>
	</entry>
</feed>