From 3415202921a92142efd97388bfab261201a34144 Mon Sep 17 00:00:00 2001
From: Ouassim Abdelmalek Ghribi <oag@big-mama.io>
Date: Fri, 31 May 2024 11:25:27 +0100
Subject: [PATCH] chore: removed all <Bleed> tags from iframes to fix videos
 overflow over left sidebar

---
 pages/introduction/basics.de.mdx         | 12 +++----
 pages/introduction/basics.en.mdx         | 12 +++----
 pages/introduction/elements.de.mdx       | 12 +++----
 pages/introduction/elements.en.mdx       | 12 +++----
 pages/introduction/examples.en.mdx       | 12 +++----
 pages/introduction/settings.de.mdx       | 12 +++----
 pages/introduction/settings.en.mdx       | 12 +++----
 pages/introduction/tips.de.mdx           | 12 +++----
 pages/introduction/tips.en.mdx           | 12 +++----
 pages/models/llama-3.de.mdx              | 12 +++----
 pages/models/llama-3.en.mdx              | 12 +++----
 pages/models/sora.de.mdx                 | 42 ++++++++++--------------
 pages/models/sora.en.mdx                 | 42 ++++++++++--------------
 pages/models/sora.jp.mdx                 | 42 ++++++++++--------------
 pages/research/guided-cot.en.mdx         | 14 ++++----
 pages/research/infini-attention.de.mdx   | 14 ++++----
 pages/research/infini-attention.en.mdx   | 14 ++++----
 pages/research/llm-recall.de.mdx         | 12 +++----
 pages/research/llm-recall.en.mdx         | 14 ++++----
 pages/research/rag-faithfulness.de.mdx   | 12 +++----
 pages/research/rag-faithfulness.en.mdx   | 14 ++++----
 pages/research/rag_hallucinations.en.mdx | 14 ++++----
 pages/research/synthetic_data.en.mdx     | 14 ++++----
 pages/research/thoughtsculpt.de.mdx      | 14 ++++----
 pages/research/thoughtsculpt.en.mdx      | 12 +++----
 pages/techniques/zeroshot.de.mdx         | 12 +++----
 pages/techniques/zeroshot.en.mdx         | 14 ++++----
 27 files changed, 183 insertions(+), 249 deletions(-)

diff --git a/pages/introduction/basics.de.mdx b/pages/introduction/basics.de.mdx
index 3f2fd4f..d02890c 100644
--- a/pages/introduction/basics.de.mdx
+++ b/pages/introduction/basics.de.mdx
@@ -27,13 +27,11 @@ Wenn Sie den OpenAI Playground oder einen anderen Playground verwenden, dann kö
 
 Hier ist eine Anleitung, wie man mit dem OpenAI Playground beginnen kann:
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/iwYtzPJELkk?si=irua5h_wHrkNCY0V" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/iwYtzPJELkk?si=irua5h_wHrkNCY0V" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 
 Zu beachten ist, dass man bei der Verwendung der OpenAI-Chatmodelle wie `gtp-3.5-turbo` oder `gpt-4` die Struktur des Prompts mit drei verschiedenen Rollen gestalten kann: `system`, `user` und `assistant`. Eine Eingabe mit `system` ist nicht erforderlich, hilft aber, das Gesamtverhalten des Assistenten festzulegen. Das obige Beispiel beinhaltet nur eine Nutzernachricht, mit der man das Modell direkt auffordern kann. Zur Vereinfachung wird in allen Beispielen, außer es ist ausdrücklich erwähnt, nur die `user`-Nachricht verwendet, um das `gtp-3.5-turbo` Modell zu prompten. Die `assistant`-Nachricht im obigen Beispiel entspricht der Modellantwort. Man kann auch eine Assistentennachricht definieren, um Beispiele für das gewünschte Verhalten zu übermitteln, das man erreichen möchte. Mehr über das Arbeiten mit Chatmodellen kann man [hier](https://www.promptingguide.ai/models/chatgpt) erfahren.
diff --git a/pages/introduction/basics.en.mdx b/pages/introduction/basics.en.mdx
index 72d58a3..effb162 100644
--- a/pages/introduction/basics.en.mdx
+++ b/pages/introduction/basics.en.mdx
@@ -27,13 +27,11 @@ If you are using the OpenAI Playground or any other LLM playground, you can prom
 
 Here is a tutorial on how to get started with the OpenAI Playground:
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/iwYtzPJELkk?si=irua5h_wHrkNCY0V" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/iwYtzPJELkk?si=irua5h_wHrkNCY0V" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 
 Something to note is that when using the OpenAI chat models like `gpt-3.5-turbo` or `gpt-4`, you can structure your prompt using three different roles: `system`, `user`, and `assistant`. The system message is not required but helps to set the overall behavior of the assistant. The example above only includes a user message which you can use to directly prompt the model. For simplicity, all of the examples, except when it's explicitly mentioned, will use only the `user` message to prompt the `gpt-3.5-turbo` model. The `assistant` message in the example above corresponds to the model response. You can also define an assistant message to pass examples of the desired behavior you want. You can learn more about working with chat models [here](https://www.promptingguide.ai/models/chatgpt).
diff --git a/pages/introduction/elements.de.mdx b/pages/introduction/elements.de.mdx
index 0281188..2910e8c 100644
--- a/pages/introduction/elements.de.mdx
+++ b/pages/introduction/elements.de.mdx
@@ -14,13 +14,11 @@ Ein Prompt enthält eines oder mehrere der folgenden Elemente:
 
 **Ausgabeindikator** - die Art oder das Format der Ausgabe.
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/kgBZhJnh-vk?si=-a-KvhmXFJMtAuCB" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/kgBZhJnh-vk?si=-a-KvhmXFJMtAuCB" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Um die Elemente eines Prompts besser zu demonstrieren, hier ein einfaches Beispiel, das darauf abzielt, eine Textklassifizierungsaufgabe durchzuführen:
 
diff --git a/pages/introduction/elements.en.mdx b/pages/introduction/elements.en.mdx
index eaeb840..1d21fff 100644
--- a/pages/introduction/elements.en.mdx
+++ b/pages/introduction/elements.en.mdx
@@ -14,13 +14,11 @@ A prompt contains any of the following elements:
 
 **Output Indicator** - the type or format of the output.
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/kgBZhJnh-vk?si=-a-KvhmXFJMtAuCB" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/kgBZhJnh-vk?si=-a-KvhmXFJMtAuCB" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 To demonstrate the prompt elements better, here is a simple prompt that aims to perform a text classification task:
 
diff --git a/pages/introduction/examples.en.mdx b/pages/introduction/examples.en.mdx
index 967e309..e72905b 100644
--- a/pages/introduction/examples.en.mdx
+++ b/pages/introduction/examples.en.mdx
@@ -19,13 +19,11 @@ Topics:
 
 ---
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/TBhRC4Dath4?si=6nwh0GuYAOv1H6yT" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/TBhRC4Dath4?si=6nwh0GuYAOv1H6yT" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 ## Text Summarization
 One of the standard tasks in natural language generation is text summarization. Text summarization can include many different flavors and domains. In fact, one of the most promising applications of language models is the ability to summarize articles and concepts into quick and easy-to-read summaries. Let's try a basic summarization task using prompts.
diff --git a/pages/introduction/settings.de.mdx b/pages/introduction/settings.de.mdx
index 270d5fc..70e02e0 100644
--- a/pages/introduction/settings.de.mdx
+++ b/pages/introduction/settings.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/CB0H7esOl68?si=OECAnvgnvJHy0qZ2" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/CB0H7esOl68?si=OECAnvgnvJHy0qZ2" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Beim Entwerfen und Testen von Prompts interagieren Sie normalerweise über eine API mit dem LLM. Sie können einige Parameter konfigurieren, um unterschiedliche Ergebnisse für Ihre Prompts zu erhalten. Das Anpassen dieser Einstellungen ist wichtig, um die Zuverlässigkeit und Erwünschtheit der Antworten zu verbessern, und es bedarf des Experimentierens, um die richtigen Einstellungen für Ihre Anwendungsfälle herauszufinden. Unten finden Sie die gängigen Einstellungen, auf die Sie bei der Verwendung verschiedener LLM-Anbieter stoßen werden:
 
diff --git a/pages/introduction/settings.en.mdx b/pages/introduction/settings.en.mdx
index 1191a45..8009599 100644
--- a/pages/introduction/settings.en.mdx
+++ b/pages/introduction/settings.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/CB0H7esOl68?si=OECAnvgnvJHy0qZ2" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/CB0H7esOl68?si=OECAnvgnvJHy0qZ2" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 When designing and testing prompts, you typically interact with the LLM via an API. You can configure a few parameters to get different results for your prompts. Tweaking these settings are important to improve reliability and desirability of responses and it takes  a bit of experimentation to figure out the proper settings for your use cases. Below are the common settings you will come across when using different LLM providers:
 
diff --git a/pages/introduction/tips.de.mdx b/pages/introduction/tips.de.mdx
index 9e0594b..36a7f0a 100644
--- a/pages/introduction/tips.de.mdx
+++ b/pages/introduction/tips.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/7M6CSCIMJ3k?si=BgaVt9g1vS4BQzXZ" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/7M6CSCIMJ3k?si=BgaVt9g1vS4BQzXZ" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Hier sind einige Tipps, die Sie beim Entwerfen Ihrer Prompts im Kopf behalten sollten:
 
diff --git a/pages/introduction/tips.en.mdx b/pages/introduction/tips.en.mdx
index b2b4c58..3dd686b 100644
--- a/pages/introduction/tips.en.mdx
+++ b/pages/introduction/tips.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/7M6CSCIMJ3k?si=BgaVt9g1vS4BQzXZ" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/7M6CSCIMJ3k?si=BgaVt9g1vS4BQzXZ" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Here are some tips to keep in mind while you are designing your prompts:
 
diff --git a/pages/models/llama-3.de.mdx b/pages/models/llama-3.de.mdx
index 8315c59..6dfa38b 100644
--- a/pages/models/llama-3.de.mdx
+++ b/pages/models/llama-3.de.mdx
@@ -40,10 +40,8 @@ Die Lizenzinformationen für die Llama 3 Modelle können auf der [Modellkarte](h
 
 Hier folgt eine längere Bewertung von Llama 3:
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/h2aEmciRd6U?si=m7-xXu5IWpB-6mE0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/h2aEmciRd6U?si=m7-xXu5IWpB-6mE0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
diff --git a/pages/models/llama-3.en.mdx b/pages/models/llama-3.en.mdx
index 3d57665..ff9649f 100644
--- a/pages/models/llama-3.en.mdx
+++ b/pages/models/llama-3.en.mdx
@@ -40,10 +40,8 @@ The licensing information for the Llama 3 models can be found on the [model card
 
 Here is a longer review of Llama 3:
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/h2aEmciRd6U?si=m7-xXu5IWpB-6mE0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
\ No newline at end of file
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/h2aEmciRd6U?si=m7-xXu5IWpB-6mE0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
diff --git a/pages/models/sora.de.mdx b/pages/models/sora.de.mdx
index 3f1644b..6962933 100644
--- a/pages/models/sora.de.mdx
+++ b/pages/models/sora.de.mdx
@@ -15,14 +15,12 @@ Prompt:
 Eine stilvolle Frau geht eine Tokioter Straße entlang, die von warm leuchtenden Neonlichtern und animierter Stadtschilderung erfüllt ist. Sie trägt eine schwarze Lederjacke, ein langes rotes Kleid und schwarze Stiefel und hat eine schwarze Handtasche dabei. Sie trägt Sonnenbrille und roten Lippenstift. Sie geht selbstbewusst und locker. Die Straße ist feucht und spiegelnd, wodurch ein Spiegeleffekt der bunten Lichter entsteht. Viele Fußgänger gehen umher.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 Prompt:
 
@@ -30,14 +28,12 @@ Prompt:
 Ein Filmtrailer, der die Abenteuer des 30-jährigen Raumfahrers zeigt, der einen roten Woll- gestrickten Motorradhelm trägt, blauer Himmel, Salzwüste, kinematografischer Stil, auf 35mm Film gedreht, leuchtende Farben.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 *Videoquelle: https://openai.com/sora*
 
@@ -54,14 +50,12 @@ Prompt:
 Szene im Schritt-Druck-Verfahren einer rennenden Person, kinematografische Filmaufnahme in 35mm.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 *Videoquelle: https://openai.com/sora*
 
diff --git a/pages/models/sora.en.mdx b/pages/models/sora.en.mdx
index 1058e72..a77327f 100644
--- a/pages/models/sora.en.mdx
+++ b/pages/models/sora.en.mdx
@@ -15,14 +15,12 @@ Prompt:
 A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 Prompt:
 
@@ -30,14 +28,12 @@ Prompt:
 A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 *Video source: https://openai.com/sora*
 
@@ -54,14 +50,12 @@ Prompt:
 Prompt: Step-printing scene of a person running, cinematic film shot in 35mm.
 ```
 
-<Bleed >
-  <iframe
-    src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 *Video source: https://openai.com/sora*
 
diff --git a/pages/models/sora.jp.mdx b/pages/models/sora.jp.mdx
index 045ac7b..a25a6a8 100644
--- a/pages/models/sora.jp.mdx
+++ b/pages/models/sora.jp.mdx
@@ -16,14 +16,12 @@ Prompt:
 スタイリッシュな女性が、暖かく光るネオンとアニメーションの街の看板で満ちた東京の通りを歩いています。彼女は黒のレザージャケット、長い赤いドレス、黒いブーツを着用し、黒いハンドバッグを持っています。サングラスと赤いリップスティックを身につけています。彼女は自信を持って、そしてカジュアルに歩きます。通りは湿っており、反射して、カラフルな光のミラー効果を生み出しています。多くの歩行者が歩いています。
 ```
 
-<Bleed>
-  <iframe
-    src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/tokyo-walk.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 Prompt:
 
@@ -31,14 +29,12 @@ Prompt:
 30歳の宇宙飛行士の冒険を描いた映画の予告編で、赤いウールの編み込みモーターサイクルヘルメットを着用し、青空、塩の砂漠、シネマティックスタイル、35mmフィルムで撮影され、鮮やかな色彩。
 ```
 
-<Bleed>
-  <iframe
-    src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/mitten-astronaut.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 _Video source: https://openai.com/sora_
 
@@ -56,14 +52,12 @@ Prompt:
 プロンプト：35mmの映画フィルムで撮影された、走る人物のステッププリントシーン。
 ```
 
-<Bleed>
-  <iframe
-    src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
-    width="100%"
-    height="300px"
-    title="SWR-States"
-  />
-</Bleed>
+<iframe
+  src="https://cdn.openai.com/sora/videos/backward-jogger.mp4"
+  width="100%"
+  height="300px"
+  title="SWR-States"
+/>
 
 _Video source: https://openai.com/sora_
 
diff --git a/pages/research/guided-cot.en.mdx b/pages/research/guided-cot.en.mdx
index fd1f91c..16dc934 100644
--- a/pages/research/guided-cot.en.mdx
+++ b/pages/research/guided-cot.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/O3bl0qURONM?si=Hwdc_o0qHpw8QRsY" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/O3bl0qURONM?si=Hwdc_o0qHpw8QRsY" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 A new paper by [Lee et al. (2024)](https://arxiv.org/abs/2404.03414) proposes to improve reasoning in LLMs using small language models.
 
@@ -25,4 +23,4 @@ The framework is tested on multi-hop extractive question answering and outperfor
 
 The LM-guided CoT prompting approach proposed in this paper outperforms both standard prompting and CoT prompting. Self-consistency decoding also enhances performance.   
 
-This approach shows a clever use of small language models for rationale generation. The results are remarkable given that larger language models are preferred for this capability over smaller ones. Decomposing tasks in this way is something developers should think deeply about. Not everything needs to be done by the large models. When fine-tuning, it's useful to think about what exact aspect you want to optimize and test to see if a small language model can do it for you.
\ No newline at end of file
+This approach shows a clever use of small language models for rationale generation. The results are remarkable given that larger language models are preferred for this capability over smaller ones. Decomposing tasks in this way is something developers should think deeply about. Not everything needs to be done by the large models. When fine-tuning, it's useful to think about what exact aspect you want to optimize and test to see if a small language model can do it for you.
diff --git a/pages/research/infini-attention.de.mdx b/pages/research/infini-attention.de.mdx
index 8505dce..347353c 100644
--- a/pages/research/infini-attention.de.mdx
+++ b/pages/research/infini-attention.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/tOaTaQ8ZGRo?si=pFP-KiLe63Ppl9Pd" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/tOaTaQ8ZGRo?si=pFP-KiLe63Ppl9Pd" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Ein neues [Paper](https://arxiv.org/abs/2404.07143) von Google integriert kompressiven Speicher in eine Vanilla Dot-Product Attention-Schicht.
 
@@ -24,4 +22,4 @@ Dieser Ansatz übertrifft Basismodelle beim langkontextuellen Sprachmodellieren
 
 Sie zeigen auch, dass ein 1B LLM natürlich auf eine Sequenzlänge von 1M skaliert werden kann und ein 8B-Modell ein neues SoTA-Ergebnis bei einer Buchzusammenfassungsaufgabe mit einer Länge von 500K erreicht.
 
-Angesichts der wachsenden Bedeutung von langkontextuellen LLMs könnte ein effektives Speichersystem leistungsstarke Fähigkeiten im Bereich des Schlussfolgerns, Planens, der kontinuierlichen Anpassung und bisher in LLMs nicht gesehene Fähigkeiten freisetzen.
\ No newline at end of file
+Angesichts der wachsenden Bedeutung von langkontextuellen LLMs könnte ein effektives Speichersystem leistungsstarke Fähigkeiten im Bereich des Schlussfolgerns, Planens, der kontinuierlichen Anpassung und bisher in LLMs nicht gesehene Fähigkeiten freisetzen.
diff --git a/pages/research/infini-attention.en.mdx b/pages/research/infini-attention.en.mdx
index 035925e..4853aa9 100644
--- a/pages/research/infini-attention.en.mdx
+++ b/pages/research/infini-attention.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/tOaTaQ8ZGRo?si=pFP-KiLe63Ppl9Pd" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/tOaTaQ8ZGRo?si=pFP-KiLe63Ppl9Pd" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 A new [paper](https://arxiv.org/abs/2404.07143) by Google integrates compressive memory into a vanilla dot-product attention layer. 
 
@@ -24,4 +22,4 @@ This approach outperforms baseline models on long-context language modeling with
 
 They also show that a 1B LLM can naturally scale to a 1M sequence length and a 8B model achieves a new SoTA result on a 500K length book summarization task.
 
-Given how important long-context LLMs are becoming having an effective memory system could unlock powerful reasoning, planning, continual adaption, and capabilities not seen before in LLMs.
\ No newline at end of file
+Given how important long-context LLMs are becoming having an effective memory system could unlock powerful reasoning, planning, continual adaption, and capabilities not seen before in LLMs.
diff --git a/pages/research/llm-recall.de.mdx b/pages/research/llm-recall.de.mdx
index 5ea0226..5e71741 100644
--- a/pages/research/llm-recall.de.mdx
+++ b/pages/research/llm-recall.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/2cNO76lIZ4s?si=tbbdo-vnr56YQ077" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/2cNO76lIZ4s?si=tbbdo-vnr56YQ077" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Dieses neue [Paper von Machlab und Battle (2024)](https://arxiv.org/abs/2404.08865) analysiert die In-Context Recall-Leistung verschiedener LLMs anhand mehrerer Nadel-im-Heuhaufen-Tests.
 
diff --git a/pages/research/llm-recall.en.mdx b/pages/research/llm-recall.en.mdx
index 7dd8da4..a1e83bb 100644
--- a/pages/research/llm-recall.en.mdx
+++ b/pages/research/llm-recall.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/2cNO76lIZ4s?si=tbbdo-vnr56YQ077" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/2cNO76lIZ4s?si=tbbdo-vnr56YQ077" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 This new [paper by Machlab and Battle (2024)](https://arxiv.org/abs/2404.08865) analyzes the in-context recall performance of different LLMs using several needle-in-a-haystack tests.
 
@@ -24,4 +22,4 @@ The recall ability of a model can be improved with increasing size, enhancing th
 
 Important practical tip from the paper: "Continued evaluation will further inform the selection of LLMs for individual use cases, maximizing their impact and efficiency in real-world applications as the technology continues to evolve."
 
-The takeaways from this paper are the importance of careful prompt design, establishing a continuous evaluation protocol, and testing different model enhancement strategies to improve recall and utility.
\ No newline at end of file
+The takeaways from this paper are the importance of careful prompt design, establishing a continuous evaluation protocol, and testing different model enhancement strategies to improve recall and utility.
diff --git a/pages/research/rag-faithfulness.de.mdx b/pages/research/rag-faithfulness.de.mdx
index 04fd4b9..4c1457c 100644
--- a/pages/research/rag-faithfulness.de.mdx
+++ b/pages/research/rag-faithfulness.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/eEU1dWVE8QQ?si=b-qgCU8nibBCSX8H" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/eEU1dWVE8QQ?si=b-qgCU8nibBCSX8H" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Dieses neue Paper von [Wu et al. (2024)](https://arxiv.org/abs/2404.10198) zielt darauf ab, das Kräftemessen zwischen den RAG-Modellen und der internen Priorisierung von LLMs zu quantifizieren.
 
diff --git a/pages/research/rag-faithfulness.en.mdx b/pages/research/rag-faithfulness.en.mdx
index 143ecee..2a2b91f 100644
--- a/pages/research/rag-faithfulness.en.mdx
+++ b/pages/research/rag-faithfulness.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/eEU1dWVE8QQ?si=b-qgCU8nibBCSX8H" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/eEU1dWVE8QQ?si=b-qgCU8nibBCSX8H" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 This new paper by [Wu et al. (2024)](https://arxiv.org/abs/2404.10198) aims to quantify the tug-of-war between RAG and LLMs' internal prior.  
 
@@ -23,4 +21,4 @@ When the documents contain more incorrect values and the LLM's internal prior is
 
 The paper also reports that "the more the modified information deviates from the model's prior, the less likely the model is to prefer it."  
 
-So many developers and companies are using RAG systems in production. This work highlights the importance of assessing risks when using LLMs given different kinds of contextual information that may contain supporting, contradicting, or completely incorrection information.
\ No newline at end of file
+So many developers and companies are using RAG systems in production. This work highlights the importance of assessing risks when using LLMs given different kinds of contextual information that may contain supporting, contradicting, or completely incorrection information.
diff --git a/pages/research/rag_hallucinations.en.mdx b/pages/research/rag_hallucinations.en.mdx
index 9219157..7f7a06f 100644
--- a/pages/research/rag_hallucinations.en.mdx
+++ b/pages/research/rag_hallucinations.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/TUL5guqZejw?si=Doc7lzyAY-SKr21L" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/TUL5guqZejw?si=Doc7lzyAY-SKr21L" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Researchers at ServiceNow shared a [new paper](https://arxiv.org/abs/2404.08189) where they discuss how to deploy an efficient RAG system for structured output tasks.
 
@@ -18,4 +16,4 @@ The RAG system combines a small language model with a very small retriever. It s
 
 The paper covers the very useful enterprise application of translating natural language requirements to workflows (formatted in JSON). So much productivity can come from this task but there is a lot of optimization that can be further achieved (eg., using speculative decoding or using YAML instead of JSON).
 
-The paper provides some great insights and practical tips on how to effectively develop RAG systems for the real world.
\ No newline at end of file
+The paper provides some great insights and practical tips on how to effectively develop RAG systems for the real world.
diff --git a/pages/research/synthetic_data.en.mdx b/pages/research/synthetic_data.en.mdx
index 58ada3d..7692a4d 100644
--- a/pages/research/synthetic_data.en.mdx
+++ b/pages/research/synthetic_data.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/YnlArBZJHY8?si=ZH3hFzwixUopxU5Z" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/YnlArBZJHY8?si=ZH3hFzwixUopxU5Z" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 This [paper](https://arxiv.org/abs/2404.07503) provides an overview of best practices and lessons learned on synthetic data for language models ans was published by Google DeepMind and other collaborators. 
 
@@ -18,4 +16,4 @@ We know for sure that the more high-quality data we give these models, the bette
 
 The paper also discusses important topics when working with synthetic data such as ensuring quality, factuality, fidelity, unbiasedness, trustworthiness, privacy, and more.
 
-There are a lot of great references mentioned in the related work section as well.
\ No newline at end of file
+There are a lot of great references mentioned in the related work section as well.
diff --git a/pages/research/thoughtsculpt.de.mdx b/pages/research/thoughtsculpt.de.mdx
index a2099dc..0f81d2f 100644
--- a/pages/research/thoughtsculpt.de.mdx
+++ b/pages/research/thoughtsculpt.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/13fr5m6ezOM?si=DH3XYfzbMsg9aeIx" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/13fr5m6ezOM?si=DH3XYfzbMsg9aeIx" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Diese Arbeit von [Chi et al. (2024)](https://arxiv.org/abs/2404.05966) stellt einen Ansatz für allgemeines Überlegen und Suchen bei Aufgaben vor, die in Komponenten zerlegt werden können.
 
@@ -24,4 +22,4 @@ Schließlich simuliert der Entscheidungssimulator (der als Teil des MCTS-Prozess
 
 Aufgrund seiner Fähigkeit zur kontinuierlichen Gedankeniteration eignet sich THOUGHTSCULPT besonders für Aufgaben wie offene Generierung, mehrstufiges Überlegen und kreative Ideenfindung.
 
-Wir könnten in Zukunft fortschrittlichere Ansätze sehen, die ähnliche Konzepte und Suchalgorithmen verwenden, um die Überlegungsfähigkeiten von LLMs zu erhöhen und die Fähigkeit, Probleme zu lösen, die komplexes Überlegen und Planen erfordern. Ein großartiges Paper, um diesen Forschungstrend im Auge zu behalten.
\ No newline at end of file
+Wir könnten in Zukunft fortschrittlichere Ansätze sehen, die ähnliche Konzepte und Suchalgorithmen verwenden, um die Überlegungsfähigkeiten von LLMs zu erhöhen und die Fähigkeit, Probleme zu lösen, die komplexes Überlegen und Planen erfordern. Ein großartiges Paper, um diesen Forschungstrend im Auge zu behalten.
diff --git a/pages/research/thoughtsculpt.en.mdx b/pages/research/thoughtsculpt.en.mdx
index cff4323..a3fb681 100644
--- a/pages/research/thoughtsculpt.en.mdx
+++ b/pages/research/thoughtsculpt.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/13fr5m6ezOM?si=DH3XYfzbMsg9aeIx" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/13fr5m6ezOM?si=DH3XYfzbMsg9aeIx" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 This work by [Chi et al. (2024)](https://arxiv.org/abs/2404.05966) presents an approach for general reasoning and search on tasks that can be decomposed into components. 
 
diff --git a/pages/techniques/zeroshot.de.mdx b/pages/techniques/zeroshot.de.mdx
index a9187ce..8069d3a 100644
--- a/pages/techniques/zeroshot.de.mdx
+++ b/pages/techniques/zeroshot.de.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/ZTaHqdkxUMs?si=EDLjgAxuFxFcrSM3" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/ZTaHqdkxUMs?si=EDLjgAxuFxFcrSM3" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Große LLMs (Language-Modelle) wie beispielsweise GPT-3.5 Turbo, GPT-4 und Claude 3 sind heute darauf abgestimmt, Anweisungen zu befolgen, und wurden mit großen Datenmengen trainiert. Groß angelegtes Training ermöglicht es diesen Modellen, einige Aufgaben auf
  "Zero-Shot"-Weise auszuführen. Zero-Shot-Prompting bedeutet, dass der Prompt, der verwendet wird, um mit dem Modell zu interagieren, keine Beispiele oder Demonstrationen enthält. Der Zero-Shot-Prompt instruiert das Modell direkt, eine Aufgabe ohne zusätzliche Beispiele auszuführen, um es zu lenken.
diff --git a/pages/techniques/zeroshot.en.mdx b/pages/techniques/zeroshot.en.mdx
index 56bb97e..b121890 100644
--- a/pages/techniques/zeroshot.en.mdx
+++ b/pages/techniques/zeroshot.en.mdx
@@ -2,13 +2,11 @@
 
 import {Bleed} from 'nextra-theme-docs'
 
-<Bleed>
-  <iframe width="100%"
-    height="415px"
-    src="https://www.youtube.com/embed/ZTaHqdkxUMs?si=EDLjgAxuFxFcrSM3" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
-    allowFullScreen
-    />
-</Bleed>
+<iframe width="100%"
+  height="415px"
+  src="https://www.youtube.com/embed/ZTaHqdkxUMs?si=EDLjgAxuFxFcrSM3" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
+  allowFullScreen
+  />
 
 Large language models (LLMs) today, such as GPT-3.5 Turbo, GPT-4, and Claude 3, are tuned to follow instructions and are trained on large amounts of data. Large-scale training makes these models capable of performing some tasks in a "zero-shot" manner. Zero-shot prompting means that the prompt used to interact with the model won't contain examples or demonstrations. The zero-shot prompt directly instructs the model to perform a task without any additional examples to steer it.
 
@@ -31,4 +29,4 @@ Note that in the prompt above we didn't provide the model with any examples of t
 
 Instruction tuning has been shown to improve zero-shot learning [Wei et al. (2022)](https://arxiv.org/pdf/2109.01652.pdf). Instruction tuning is essentially the concept of finetuning models on datasets described via instructions. Furthermore, [RLHF](https://arxiv.org/abs/1706.03741) (reinforcement learning from human feedback) has been adopted to scale instruction tuning wherein the model is aligned to better fit human preferences. This recent development powers models like ChatGPT. We will discuss all these approaches and methods in upcoming sections.
 
-When zero-shot doesn't work, it's recommended to provide demonstrations or examples in the prompt which leads to few-shot prompting. In the next section, we demonstrate few-shot prompting.
\ No newline at end of file
+When zero-shot doesn't work, it's recommended to provide demonstrations or examples in the prompt which leads to few-shot prompting. In the next section, we demonstrate few-shot prompting.