{
  "docId": "019dd923-5e88-73ef-bd5c-fd5012384ae3",
  "docSlug": "bf350dd574c19997",
  "documentTitle": "2021 Air Street Capital The State of AI Report 2021",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 38,
  "pageCount": 188,
  "prevPage": 37,
  "nextPage": 39,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 2,
  "notes": "The slide explains the CLIP architecture using a three-step process diagram.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "process_diagram"
  ],
  "metadataConfidence": 0.95,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-fd5012384ae3/38",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3#slide-38",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "CLIP is a good zero-shot learner. It performs as well as the original fully supervised ResNet-50, and, on average, it outperforms all existing models in zero-shot prediction across 27 datasets.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d28-279e610d8fc3",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.6,
        "w": 0.25,
        "x": 0.72,
        "y": 0.28
      },
      "kind": "image",
      "text": "Diagram showing 1. Contrastive pre-training, 2. Create dataset classifier from label text, 3. Use for zero-shot prediction",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "92656dc6-d22e-41aa-bca8-b0f4656ee062",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.55,
        "w": 0.65,
        "x": 0.03,
        "y": 0.31
      },
      "kind": "list",
      "text": "CLIP's powerful learned representations result from using 3 ingredients: a Vision Transformer, a contrastive objective (inspired by ConvIRT), and... scale.\nDuring contrastive pre-training, the model learns to associate each image in a batch with its text companion, while dissociating it from the other text snippets.\nTo use CLIP on a specific classification task, one needs to use prompts, where the labels of the task's dataset are reformulated to resemble the pre-training set while communicating the underlying context of the task. CLIP then predicts, among all the encoded prompts, the one which has minimal contrastive loss with the encoded image.\nCLIP is a good zero-shot learner. It performs as well as the original fully supervised ResNet-50, and, on average, it outperforms all existing models in zero-shot prediction across 27 datasets on object classification, OCR, activity recognition in videos, and geo-localization.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d6923aef-fe93-4ef5-bbd1-6cd235fb9617",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.65,
        "x": 0.03,
        "y": 0.21
      },
      "kind": "paragraph",
      "text": "OpenAI's CLIP uses 400M text-image pairs to learn image and text representations. It exhibits a solid performance across a wide variety of datasets without any fine-tuning.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "2eb40cee-e689-44ff-8b6e-89acf61380ab",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.65,
        "x": 0.03,
        "y": 0.13
      },
      "kind": "title",
      "text": "Multimodal self-supervision plus scale equals a powerful representer",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a5f9e25b-324b-4225-9f50-0896a245cd04",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "2x2 matrix",
      "slug": "matrix-2x2",
      "agent": null,
      "layer": "slide",
      "matchId": "eb8e43f0-6436-47eb-8e8e-5c16dce33180",
      "evidence": "The slide implies a 2x2 matrix or similar structure in presenting the ingredients for CLIP's success.",
      "confidence": 0.6
    },
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8ed0-faef7ab54f62",
      "evidence": "Title 'Multimodal self-supervision plus scale equals a powerful representer'.",
      "confidence": 80
    },
    {
      "name": "Concrete Language",
      "slug": "concrete-language",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8ed0-fe83dd255e24",
      "evidence": "27 datasets, ResNet-50 baseline, 400M pairs.",
      "confidence": 70
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 152,
      "from": 5,
      "beatId": "019dd95a-0682-776c-8e35-0affbadec38e",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Definitions p.5-6, Exec Summary p.7, then Sections 1-3 catalog evidence.",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 152,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e35-1b3ee3a1c012",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Rising Action",
      "beatSlug": "mountain-rising-action",
      "evidence": "Three sections accumulating evidence of accelerating progress.",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 40,
      "from": 36,
      "name": "Aha Moment",
      "slug": "03-aha-moment",
      "bestFor": "Data-heavy sections, research findings, analytical arguments",
      "matchId": "019dd95a-07fe-70ce-8d3d-53192fc75199",
      "evidence": "Diffusion models > GANs, then CLIP, then DALL-E payoff.",
      "position": 5,
      "objective": "Reveal multimodal generative AI breakthrough",
      "structure": "The Problem/Question -> What the Data Shows -> The Insight",
      "confidence": 72,
      "description": "Establish a tension, present data, then deliver the insight that changes everything"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}