{
  "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": 29,
  "pageCount": 188,
  "prevPage": 28,
  "nextPage": 30,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 20,
  "notes": "The slide compares DreamerV2 against IQN, Rainbow, C51, and DQN across four performance metrics (Gamer Median, Gamer Mean, Record Mean, Clipped Record Mean) over 200 million steps.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "line_chart"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-fd5012384ae3/29",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3#slide-29",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "DreamerV2 is the first model-based RL agent trained on a single GPU to surpass human level performance on 55 popular tasks of the Atari benchmark.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d26-595cc33b96c0",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.35,
        "w": 0.78,
        "x": 0.12,
        "y": 0.52
      },
      "kind": "chart",
      "text": "Atari performance over 200 million steps",
      "attrs": null,
      "subkind": "line",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "bc6ddfd1-f9dd-4b60-b176-0d92fda70c63",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.65,
        "x": 0.18,
        "y": 0.83
      },
      "kind": "legend",
      "text": "DreamerV2, IQN, Rainbow, C51, DQN",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "e9a7a2b8-3837-4dca-b3d7-e80b050d3a24",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.06,
        "w": 0.85,
        "x": 0.04,
        "y": 0.36
      },
      "kind": "list",
      "text": "DreamerV2 vastly outperforms other RL agents trained with the same computational budget, across all performance aggregation metrics.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "fce5c106-987e-4830-b993-4f1b13026289",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Atari benchmark performance",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d26-5d031b2fd5af",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.92,
        "x": 0.04,
        "y": 0.21
      },
      "kind": "paragraph",
      "text": "DreamerV2 is the first model-based RL agent trained on a single GPU to surpass human level performance on 55 popular tasks of the Atari benchmark. The agent learns behaviors purely within the latent space of a world model trained from pixels, which makes these behaviors more generalisable to solving future tasks more efficiently.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "8dc6e954-fe2b-405f-b5c8-6495de17c6c0",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.45,
        "x": 0.04,
        "y": 0.14
      },
      "kind": "title",
      "text": "Superhuman world models for Atari, but on a budget",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "75f47781-c7f9-4e2a-952f-7a52aec5159f",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Audience Definition",
      "slug": "audience-definition",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "732456b0-8989-4330-aca6-8084656497a3",
      "evidence": "The slide presents a case study on DreamerV2, indicating a specific audience in mind for the technical details.",
      "confidence": 0.7
    }
  ],
  "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": 32,
      "from": 28,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3d-4c5c6602de39",
      "evidence": "MuZero, DreamerV2, generalisation, Leela coaching, then statistics rigour critique.",
      "position": 4,
      "objective": "Survey reinforcement learning advances + concerns",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 70,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}