{
  "docId": "019dd923-5de0-76bd-a168-6159b949e2c2",
  "docSlug": "ff5d549e5027972a",
  "documentTitle": "GPT-3 and the actuarial landscape",
  "authorId": "OliverWyman",
  "authorName": "Oliver Wyman",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "strategy_consulting",
  "sourceTypeLabel": "Strategy consulting",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.778,
  "pageNumber": 26,
  "pageCount": 46,
  "prevPage": 25,
  "nextPage": 27,
  "slideType": "other",
  "function": "present_framework",
  "density": "dense",
  "nDataPoints": 0,
  "notes": "The slide uses a simple linguistic example to explain the concept of self-attention in AI models.",
  "elementsJson": [
    "headline_text",
    "paragraph",
    "bullet_list"
  ],
  "metadataConfidence": 0.9,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5de0-76bd-a168-6159b949e2c2/26",
  "deckHref": "/decks/019dd923-5de0-76bd-a168-6159b949e2c2",
  "deckJsonHref": "/decks/019dd923-5de0-76bd-a168-6159b949e2c2.json",
  "deckAnchorHref": "/decks/019dd923-5de0-76bd-a168-6159b949e2c2#slide-26",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Therefore, we want the next word predictions to highly depend on \"music\" and \"lifting\" and not so much on \"enthusiastic\" and \"apartment\".",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd951-d9f1-764a-82eb-ae535d3ded7a",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.8,
        "x": 0.037,
        "y": 0.57
      },
      "kind": "list",
      "text": "In the two sentences above:\n- The words \"music\" and \"lifting\" give a lot of meaning to the token \"heavy\", since those tokens help specify the context.\n- The words \"enthusiastic\" and \"apartment\", however are not very useful in finding out what is \"heavy\".",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "177faf72-834c-4907-a7c7-5d4491aca7f9",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.1,
        "w": 0.6,
        "x": 0.037,
        "y": 0.44
      },
      "kind": "list",
      "text": "I can enjoy almost any music genre, but I was never enthusiastic about heavy ___.\nI run instead of lifting, because my apartment building's gym doesn't have heavy ___.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f1a970d3-8803-4760-92f2-a04672197523",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.9,
        "x": 0.037,
        "y": 0.75
      },
      "kind": "paragraph",
      "text": "Therefore, we want the next word predictions to highly depend on \"music\" and \"lifting\" and not so much on \"enthusiastic\" and \"apartment\".",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "705d64e1-bc31-4e66-9084-77bc65a5595c",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.25,
        "x": 0.037,
        "y": 0.39
      },
      "kind": "paragraph",
      "text": "Motivation for Self-Attention",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "9e2197e5-1525-46f9-99ed-1e3c8de75b71",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.1,
        "w": 0.92,
        "x": 0.037,
        "y": 0.24
      },
      "kind": "paragraph",
      "text": "Self-Attention is the mechanism used by transformer models to weigh the importance of difference words in a sentence or piece of text based on their relationships to other words.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "cd896e53-b0d4-4c15-808a-cfd4a81f2a37",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.32,
        "x": 0.037,
        "y": 0.05
      },
      "kind": "title",
      "text": "ATTENTION IS ALL YOU NEED",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "de2b3eb3-0f1c-4134-b270-509b1dd885dd",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 31,
      "from": 17,
      "beatId": "5259c9cc-c0ba-4428-ac14-7a6ba8d02be1",
      "arcName": "The Sparkline",
      "arcSlug": "sparkline",
      "beatName": "What Could Be",
      "beatSlug": "sparkline-what-could-be-2",
      "evidence": "The deck explores the potential applications of GPT-3 in the actuarial field.",
      "position": 2,
      "confidence": 0.8,
      "parentBeatName": "Turn",
      "parentBeatSlug": "turn"
    }
  ],
  "loops": [],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}